142 min 2026-02
Dario Amodei — "We are near the end of the exponential"
概要
Dario Amodei 深度访谈：指数增长接近终点、编码生产力频谱、AGI 时间线一到三年、Claude Code 诞生故事、安全治理路径与地缘博弈
核心洞察
元信息
- 被访者：Dario Amodei，Anthropic CEO
- 访谈者：Dwarkesh Patel
- 来源：The Lunar Society Podcast
- 时长：142 分钟
Executive Summary
- 技术指数仍在按预期推进，但公众对"指数即将走完"这件事的无感令 Dario 感到震惊。 他给出明确时间表——90% 的概率在 10 年内、50% 的概率在 1-3 年内到达"数据中心里的天才之国"（country of geniuses in a data center），即在数字世界中拥有一群能力匹敌诺贝尔奖得主的 AI 智能体。
- Anthropic 的营收轨迹印证了技术扩散的速度：2023 年 1 亿 → 2024 年 10 亿 → 2025 年 90-100 亿美元，2026 年 1 月单月又新增数十亿。 但 Dario 反复强调"极快但非瞬时"——经济扩散不会无限快，企业采购流程、安全合规、变更管理都在产生摩擦。
- RL 扩展正在重现预训练时代的故事：从窄域（数学竞赛）走向广域，泛化正在涌现。 "大计算团块假说"（Big Blob of Compute Hypothesis）自 2017 年提出至今未被推翻，预训练 + RL 双阶段共同构成了"介于人类进化与人类学习之间"的新型学习范式。
- 编程是 AI 渗透最深的领域：AI 已写 90% 的代码，但从"90% 代码"到"100% 端到端软件工程"再到"软件工程师需求减少 90%"是一个漫长的频谱。 Anthropic 内部编码生产力提升约 15-20%，六个月前仅 5%——雪球正在加速滚动。
- Claude Code 从内部工具到品类领导者的路径，本质是"狗粮驱动的产品-市场匹配"： Anthropic 2500 名工程师每天使用，形成了模型能力与产品迭代的飞轮。
- 在安全与治理层面，Dario 拒绝"十年联邦禁令"式的放任，主张从透明度标准起步、按风险证据逐步升级监管。 他对 AI 生物恐怖主义风险的紧迫感远超对"情感聊天机器人禁令"的担忧，并呼吁在今年晚些时候就可能采取实质性立法行动。
- 主线串联：Dario 的核心叙事是一条"陡峭但平滑的指数曲线"——技术能力飞速前进、经济扩散紧随其后但有摩擦、安全治理必须与之赛跑。他既是加速主义者（"不到 2030 年就会有万亿美元收入"），也是审慎主义者（"如果对需求预测偏差一年，就会破产"），这种张力贯穿了整场对话。
"大计算团块假说"八年未被推翻：预训练 + RL 双引擎驱动能力指数
核心要点：AI 能力的核心驱动力始终是"七大要素"——算力、数据量、数据质量与分布、训练时长、可扩展的目标函数、数值稳定性——RL 扩展只是同一规律的最新实例。
- Dario 早在 2017 年 GPT-1 刚发布时就写了"大计算团块假说"（Big Blob of Compute Hypothesis），列出七个关键要素。八年过去，他认为没有任何观察与之矛盾。Rich Sutton 两年后发表的"苦涩的教训"（The Bitter Lesson）表达了相似理念。
- RL 扩展正在复刻预训练故事：多家公司的公开数据显示，在 AIME 等数学竞赛上模型表现与训练时长呈对数线性关系。Anthropic 自身也在广泛 RL 任务上观察到同样的缩放规律。
- 预训练的泛化跃迁发生在数据从窄分布（GPT-1 的文学文本 / fanfiction）扩展到全互联网（GPT-2 的 Common Crawl / Reddit 链接抓取）时。RL 正在经历同样的过程：从数学竞赛 → 代码 → 多种任务。
- 针对 Rich Sutton"LLM 不够像人类学习"的质疑，Dario 提出精妙的"层级类比"：预训练介于人类进化与人类终身学习之间，上下文学习介于人类长期学习与短期学习之间——LLM 的学习模式并非落在人类学习谱系的某个精确点上，而是填充在各点之间。
"在距离指数终点如此之近的时候，人们还在讨论那些陈旧的热门政治话题，这在我看来简直疯狂。" —— Dario Amodei
编码生产力的漫长频谱：从"AI 写 90% 代码"到"SWE 需求减少 90%"是不同的世界
核心要点：AI 在编程领域的渗透最快、数据最清晰，但"代码行数"与"端到端生产力"之间存在被严重低估的鸿沟。
- Dario 八九个月前预测"AI 将在 3-6 个月内写 90% 的代码"，这已在 Anthropic 及许多下游用户中实现，但他强调这个标准极弱——"90% 的代码由 AI 写"和"不需要 90% 的软件工程师"完全是两个世界。
- 他给出了一个完整频谱：90% 代码 → 100% 代码 → 90% 端到端 SWE 任务（含编译、集群搭建、测试、写备忘录）→ 100% 端到端 SWE 任务 → SWE 创造新的高层任务 → SWE 需求减少 90%。他预计 1-2 年内可实现端到端 SWE，包括"设定技术方向、理解问题上下文"。
- Anthropic 内部体感：当前编码模型带来约 15-20% 全要素加速，六个月前仅约 5%。"5% 不值一提，但 15-20% 已开始成为多个因素中真正起作用的一个。"雪球在加速滚动。
- Dwarkesh 引用一项研究反驳：经验丰富的开发者使用 AI 工具后自我报告生产力提升，但实际合并代码量下降了 20%。Dario 的回应是：Anthropic 处于巨大商业压力下，"没有时间自欺欺人"，内部每几个月就通过模型发布验证真实产出。有工程师已经不写任何代码，GPU kernel 和芯片代码全交给 Claude。
"有人以为我在说我们不再需要 90% 的软件工程师了。这两件事之间隔着一整个世界。" —— Dario Amodei
"数据中心里的天才之国"：一到三年内到达，十年内 90% 确定
核心要点：Dario 给出了他迄今最明确的 AGI 时间表——50% 概率 1-3 年，90% 概率 10 年，5% 的不可约不确定性来自地缘灾难（台湾被入侵、晶圆厂被炸）等极端事件。
- 他将信心区分为强弱两层。弱版：10 年内到达"天才之国"的概率为 90%，"说 2035 年前不会发生在任何理性世界都应被视为非主流观点"。强版（直觉）：1-2 年内到达的概率约 50%。
- 唯一的根本性不确定来自"不可验证任务"——规划火星任务、做出 CRISPR 级别的科学发现、写小说。可验证任务（编码、数学）的路径非常确定，但从可验证域到不可验证域的泛化已经在发生，只是可能不完全。
- Anthropic 官方预测写入了公开文件：到 2026 年底或 2027 年初，AI 系统将具备"导航人类可用的数字界面、智力能力匹配或超越诺贝尔奖得主、与物理世界交互"的能力。
- 关于"持续学习"（continual learning）是否是瓶颈：Dario 认为可能根本不存在这个障碍——预训练泛化 + RL 泛化 + 百万 token 上下文学习可能就足够。但他也表示正在研究真正的持续学习，"有很大概率在一两年内解决"。
"如果我们真的有了'数据中心里的天才之国'，我们会知道的。这个房间里的每个人都会知道。华盛顿的每个人都会知道。我们现在还没有。" —— Dario Amodei
经济扩散：极快但非瞬时——10 倍年增长的天花板在哪里
核心要点：Dario 明确拒绝了两种极端叙事——"AI 没用论"和"递归自我改进瞬间奇点论"，坚持一条"陡峭但平滑的指数"。
- Anthropic 营收数据：2023 年 0→1 亿、2024 年 1→10 亿、2025 年 10→90-100 亿美元。2026 年 1 月单月又新增数十亿。"你以为曲线会放缓，但它没有。"不过 Dario 承认 2026 年内可能开始弯曲——毕竟 GDP 有限。
- Claude Code 是扩散速度的绝佳案例：个人开发者和 Series A 创业公司的采用速度比大型企业快好几个月。大型金融和制药公司采用速度远快于历史上任何新技术，但仍需经过法务、安全合规、领导层解释、3000 人滚动部署等环节。
- Dwarkesh 犀利挑战："扩散是 cope（自我安慰）。AI 的天然优势（秒读全部 Slack、无逆向选择问题、可复制）应该让扩散比人类招聘容易得多。全球每年给人类发 50 万亿美元工资，招人比招 AI 麻烦多了，但人类照样被雇用。"
- Dario 回应：扩散是真实的，但不是"AI 不重要"的理由。他的框架是：有一条快速指数（模型能力），下游还有一条快速指数（经济扩散）。后者比前者慢，但比历史上任何技术都快。他对"不到 2030 年就有万亿美元收入"高度确信。
"我们正试图让 Anthropic 的年收入增长 20 到 30 倍而不是 10 倍。企业客户说'这太有生产力了，我们要在通常的采购流程中走捷径'。他们的速度比我们卖普通 API 时快得多。" —— Dario Amodei
算力投资的"地狱级需求预测"：如何在万亿美元级别押注而不破产
核心要点：Dario 首次完整阐述了 AI 行业的资本支出逻辑——为什么既不能"YOLO 式"投千亿、也不能保守观望，以及"盈利"在这个行业中意味着什么。
- 行业算力增长速度：2025 年约 10-15 GW，每年 3 倍增长。2026 年 30-40 GW，2028 年 100 GW，2029 年 300 GW。每 GW 年化成本约 100-150 亿美元。到 2028-2029 年，行业级别达到数万亿美元。
- "如果对需求预测偏差一年，就会破产。"——假设 2027 年按 10 倍增长买下 1 万亿美元算力，如果增速只有 5 倍或晚了一年，"地球上没有任何对冲能救你"。所以 Anthropic 买了"能捕获相当强上行空间"的量，但不是激进的极限值。
- 盈利的本质是"需求预测误差的方向"：如果低估了需求（收入超预期），就盈利但研究算力被挤压；如果高估了需求，就亏损但有充足研究算力。2028 年盈利预期不是因为停止投资，而是因为推理业务的毛利率高于 50%、训练投入占比趋于稳定。
- 行业结构将是 3-4 家巨头的"库诺均衡"，类似云计算——进入门槛极高（资本 + 专业能力），不会完全竞争到零利润，也不会形成垄断。模型之间的差异化（Claude 擅长的领域 ≠ GPT 擅长的 ≠ Gemini 擅长的）比云服务更强。
"有些公司根本没有写下那张电子表格。他们不理解自己在承担什么风险。他们只是觉得酷就做了。" —— Dario Amodei
Claude Code 的诞生：从内部 CLI 到品类王者的意外路径
核心要点：Claude Code 并非战略规划的产物，而是 Anthropic "自己吃自己的狗粮"时自然生长出来的——内部产品-市场匹配先于外部发布。
- 2025 年初，Dario 判断"用自家模型显著加速自身研究的时机已到"，鼓励内部员工自由实验。最初叫 Claude CLI，后改名 Claude Code。
- 内部采用飞速增长，Dario 的逻辑很简单："我们有几百人的代表性样本，已经证明了产品-市场匹配。为什么不发布？"
- 编码是一个天然有利于 AI 的领域——代码库本身就是"外化的记忆支架"，模型通过读取代码库就获得了人类需要半年才能积累的上下文。这也解释了为什么编码进步如此之快，而其他经济活动可能更慢。
- 反馈飞轮：内部开发者发现"模型在 X 方面可以更好" → 烘焙到下一代模型 → 产品迭代 → 外部数百万用户反馈进一步加速。"这就是我们推出了编码模型而不是制药公司的原因——虽然我的背景是生物学。"
"Anthropic 有工程师已经不写任何代码了。GPU kernel、芯片代码，他们以前自己写，现在全交给 Claude。" —— Dario Amodei
AI 安全治理：从透明度起步，按证据升级，拒绝十年真空
核心要点：Dario 的监管哲学是"起步轻、升级快"——先透明度标准，再按风险证据逐步加码，关键是保持灵活性而非一步到位。
- 他明确反对美国国会提出的"十年联邦禁止州级 AI 立法"的法案：没有联邦替代方案、没有实际立法提案，在 10 年这个"永恒"的时间跨度内放任不管，面对生物武器和自主性风险，"这是疯狂的"。
- 但他也反对田纳西州那样的"禁止 AI 提供情感支持"的法案——"立法者显然对 AI 能做什么和不能做什么毫无了解。"
- 他支持的路径：联邦政府制定统一标准并"抢占"州法（preemption），但前提是联邦确实在行动而非留下真空。当前阶段以透明度标准为主；一旦风险证据充分（他认为可能今年晚些时候），快速推进针对性立法（如强制要求生物分类器）。
- 对 AI 好处被监管扼杀的担忧，Dario 认为发达国家问题不大——"当有大量金钱可赚且明显是最佳替代方案时，监管体系实际上很难阻止它"。芯片出口管制的例子证明了这一点（所有人都支持，但钱太多了所以执行不力）。他更担忧的是发展中国家和 FDA 药物审批流程的瓶颈。
"那项法案已经开始显得不合时宜了——你已经看到了反弹。" —— Dario Amodei
地缘博弈：民主国家必须在"规则制定时刻"握有更强的牌
核心要点：Dario 主张对华芯片出口管制，核心逻辑不是遏制中国人民的福祉，而是防止在 AI 带来的"关键窗口期"中，威权国家获得颠覆性军事和网络优势。
- 他区分了两层风险：(1) 攻击主导（offense-dominant）世界中，一方 AI 足以摧毁一切，比核武器更危险且可能不稳定——核威慑稳定是因为确定性，而 AI 对抗存在"双方都认为自己有 90% 胜率"的不确定性；(2) 威权政府用 AI 压迫本国人民，建成高科技极权国家。
- 他的框架：AI 指数会持续推进，但会有"若干关键时刻"——某国或某联盟在网络进攻、核威慑稳定性等维度率先到达临界点。在那个时刻会有一次隐性或显性的"后 AI 世界秩序"谈判，他希望民主国家在谈判中握有更强的牌。
- 但他并不主张推翻所有威权国家——在访谈中他坦承自己在《技术的青春期》中"探索了更激进的观点"，但承认其问题。他引用中国 70-80 年代的接触政策作为正面案例：十几亿人因此变得更富裕。
- 他提出一个更具想象力的可能：能否通过技术本身创造一种"均衡态"，使威权国家无法在保持权力的同时阻止国民私人使用 AI？"我们曾寄希望于社交媒体和互联网能瓦解威权，结果没有。但如果我们带着失败的教训用不同的技术再试一次呢？"
"独裁可能会变成一种在道德上过时的政府形式——在道德上不可行。而这场危机的严重程度足以迫使我们找到另一条路。" —— Dario Amodei
AI 宪法与价值观：原则优于规则，竞争优于垄断
核心要点：Anthropic 的 AI 宪法选择"原则导向"而非"规则列表"，是因为原则能让模型行为更一致、更好泛化；在价值观设定上，Dario 设想三层迭代反馈环路。
- 规则 vs. 原则：给模型一堆"不许做 X"的规则，模型难以理解和泛化（"不许告诉人怎么短接汽车、不许说韩语"）；给模型原则——"在正常情况下执行用户任务，但拒绝危害他人的请求"——行为更一致，边缘案例覆盖更好。
- 可纠正性（corrigibility）vs. 内在动机：模型大体上应该是"可纠正的"——执行用户指令为默认态，但设有硬性底线（如不制造生物武器）。"我们不是在造一个自行去管理世界的东西。"
- 三层迭代环路：(1) Anthropic 内部迭代——训练模型、不满意就修改宪法；(2) 公司间竞争——Anthropic、Gemini、其他公司各出宪法，外部观察者比较优劣，形成软性激励；(3) 社会参与——曾与 Collective Intelligence Project 合作做公众调研，未来可能扩展到代议制政府输入（但目前立法流程太慢）。
- Dwarkesh 将第二层类比为自由意志主义者的"群岛乌托邦"——不同治理模式竞争、人们用脚投票。Dario 承认"这个愿景有可取之处，也会出现你想不到的问题"，关键是三层环路的比例组合。
"通过教模型原则而非规则，它的行为更一致，更容易覆盖边缘情况，也更可能做到用户真正想要的事。" —— Dario Amodei
机器人革命：在"天才之国"之后再加一两年
核心要点：机器人技术的突破不依赖单一路径（不一定需要"像人类一样的学习"），而是会通过多条路径汇聚——模拟环境泛化、计算机使用技能迁移、上下文学习——实现时间窗口在"天才之国"之后追加一到两年。
- Dario 预测机器人产业也将产生万亿美元级收入，但要叠加"极快但非瞬时"的额外扩散延迟。
- 他强调 AI 将同时革新两件事：机器人硬件设计（AI 比人类更擅长）和机器人控制软件。
- 计算机使用（computer use）是当前的关键瓶颈之一——OSWorld 基准从 Anthropic 首次发布时的约 15% 攀升到 65-70%，但可靠性仍需跨过一个临界点。
"机器人技术会被革命化吗？会的。大概再加一两年。" —— Dario Amodei
发展中国家不能被落下：不靠慈善，靠内生增长
核心要点：Dario 担心 AI 红利地理分布极度不均——硅谷增长 50%、世界其他地方几乎不变——这将是"一个非常糟糕的世界"。他的解法不是慈善，而是在发展中国家建立内生 AI 产业。
- 传统追赶增长的机制是"未充分利用的劳动力 + 发达国家的资本和知识"。在劳动力不再是约束因素的 AI 时代，这个机制失效。
- 他的方案：(1) 在非洲建数据中心（"只要不是中国拥有的"）；(2) 在发展中国家扶植 AI 驱动的生物技术创业公司；(3) 确保转型期人类仍有角色时，一部分这些人类在发展中国家。
- Anthropic 已经在与全球慈善家和向撒哈拉以南非洲、印度、拉丁美洲递送医疗干预的机构合作。
- 他对发达国家并不太担忧——市场机制会推动 AI 好处的扩散。真正的瓶颈是 FDA 药物审批流程：AI 将大幅加速药物发现，但审批管线"没有准备好处理如此多的涌入"。
"增长和经济价值会来得很容易。不容易的是利益分配、财富分配和政治自由。" —— Dario Amodei
领导力与公司文化：DVQ、Slack 频道与"对 2500 人说真话"
核心要点：Dario 将约 40% 的时间用于维护 Anthropic 的文化——在一个 2500 人的组织中，他选择的杠杆不是事必躬亲，而是高频率、高透明度的沟通。
- 每两周一次的全公司"DVQ"（Dario Vision Quest，他曾试图改名未遂——"听起来像我在抽仙人掌"）：3-4 页文档，涵盖内部模型进展、产品、行业动态、地缘政治，然后回答问题。大比例员工到场或远程参加。
- 他在 Slack 上有一个频道，高频写作和评论——回应内部调查中的员工关切，"直接说、承认问题、避免企业腔"。
- 他观察到其他 AI 公司随着规模扩大出现"退相干"（decoherence）——内部争斗、互相拆台。Anthropic 的差异化在于"让每个人相信其他人也是为了正确的理由在这里"。
- Daniela（Dario 的妹妹）负责公司日常运营，联合创始人团队和招聘文化共同维系凝聚力。
"有些非常关键的决定，可能就是某人走进我办公室说'Dario，你有两分钟。做 A 还是 B？'我说'我不知道，我得去吃午饭了，做 B 吧。'然后那就是有史以来最重要的决定。" —— Dario Amodei
附录：关键人/机构/产品/数据
| 类别 | 名称 | 备注 |
|------|------|------|
| 人物 | Dario Amodei | Anthropic CEO，前 OpenAI VP of Research |
| | Daniela Amodei | Anthropic 联合创始人兼总裁，负责日常运营 |
| | Dwarkesh Patel | 访谈者，The Lunar Society 主持人 |
| | Rich Sutton | 强化学习先驱，"The Bitter Lesson"作者 |
| | Alec Radford | GPT-1 核心研究者（OpenAI 时期） |
| 机构 | Anthropic | 2500 人，2025 年营收 90-100 亿美元 |
| | Collective Intelligence Project | 与 Anthropic 合作进行 AI 宪法公众调研 |
| | Gates Foundation | Dario 提及其在消灭脊髓灰质炎方面的努力 |
| 产品 / 概念 | Claude Code | Anthropic 编码代理工具，品类领导者 |
| | "大计算团块假说" | Dario 2017 年内部文档，AI 扩展核心论文 |
| | DVQ (Dario Vision Quest) | 每两周全公司通报，3-4 页 |
| | OSWorld | 计算机使用基准，Anthropic 从约 15% 升至 65-70% |
| 关键数据 | Anthropic 营收 | 2023: ~1 亿 → 2024: ~10 亿 → 2025: ~90-100 亿美元 |
| | 行业算力规模 | 2025: 10-15 GW → 每年 3x → 2029: ~300 GW |
| | 每 GW 年化成本 | ~100-150 亿美元 |
| | 编码生产力提升 | 当前 15-20%，六个月前约 5% |
| | AGI 时间线 | 90% 概率 10 年内；50% 概率 1-3 年内 |
| 文章 / 著作 | "Machines of Loving Grace" | Dario 2024 年博文，讨论 AI 如何造福人类 |
| | "The Adolescence of Technology" | Dario 2025 年 50 页长文，讨论 AI 安全与治理 |
| | "The Bitter Lesson" | Rich Sutton 2019 年文章 |
中文翻译 English Original
**Dwarkesh Patel：** 我们三年前谈过一次。在你看来，过去三年最大的变化是什么？那时候的感觉和现在最大的不同是什么？
**Dwarkesh Patel:** We talked three years ago. In your view, what has been the biggest update over the last three years? What has been the biggest difference between what it felt like then versus now?
**Dario Amodei：** 大体上来说，底层技术的指数级增长和我预期的差不多。各方面大概有一两年的偏差。我不确定自己能否准确预测代码这个特定方向的发展。但当我审视这条指数曲线时，它基本上和我预期的一样——模型从聪明的高中生水平，到聪明的大学生水平，再到开始能做博士和专业级别的工作，在代码领域甚至已经超越了这个范围。前沿的进展有些参差不齐，但大致符合我的预期。最让我惊讶的是，公众对我们距离指数增长终点有多近这件事缺乏认知。在我看来，这绝对是荒谬的——无论是圈内人还是圈外人，都还在讨论那些老生常谈的政治热点话题，而我们已经接近指数增长的终点了。
**Dario Amodei:** Broadly speaking, the exponential of the underlying technology has gone about as I expected it to go. There's plus or minus a year or two here and there. I don't know that I would've predicted the specific direction of code. But when I look at the exponential, it is roughly what I expected in terms of the march of the models from smart high school student to smart college student to beginning to do PhD and professional stuff, and in the case of code reaching beyond that. The frontier is a little bit uneven, but it's roughly what I expected. What has been the most surprising thing is the lack of public recognition of how close we are to the end of the exponential. To me, it is absolutely wild that you have people — within the bubble and outside the bubble — talking about the same tired, old hot-button political issues, when we are near the end of the exponential.
**Dwarkesh Patel：** 我想了解一下这条指数曲线现在具体是什么样子。三年前我们录制时我问你的第一个问题就是"scaling 是怎么回事，为什么有效？"现在我有一个类似的问题，但感觉更复杂了。至少从公众的角度看，三年前有许多数量级的计算量上的公开趋势，你可以看到 loss 是如何改善的。现在我们有了 RL scaling，但没有公开的 scaling law。甚至连基本的逻辑都不清楚。这到底是在教模型技能？还是在教元学习？目前的 scaling 假说到底是什么？
**Dwarkesh Patel:** I want to understand what that exponential looks like right now. The first question I asked you when we recorded three years ago was, "what's up with scaling and why does it work?" I have a similar question now, but it feels more complicated. At least from the public's point of view, three years ago there were well-known public trends across many orders of magnitude of compute where you could see how the loss improves. Now we have RL scaling and there's no publicly known scaling law for it. It's not even clear what the story is. Is this supposed to be teaching the model skills? Is it supposed to be teaching meta-learning? What is the scaling hypothesis at this point?
**Dario Amodei：** 实际上，我的假说和 2017 年时完全一样。我想上次也提到过，我写过一个文档叫"The Big Blob of Compute Hypothesis"（大团计算假说）。它不是专门关于语言模型 scaling 的。当我写那篇文档时，GPT-1 刚刚发布。那只是众多方向之一。那个年代还有机器人学。人们试图把推理作为一个独立于语言模型的东西来研究，还有 AlphaGo 和 OpenAI 的 Dota 那种 RL 的 scaling。大家还记得 DeepMind 的 StarCraft，AlphaStar。那篇文档是一个更通用的文件。Rich Sutton 几年后发表了"The Bitter Lesson"（苦涩的教训）。这个假说基本上是一样的。它说的是，所有的聪明才智、所有的技巧、所有"我们需要一个新方法来做某件事"，这些都不太重要。只有几件事是重要的。我列了七项。第一是你拥有多少原始算力。第二是数据的数量。第三是数据的质量和分布。它需要是一个广泛的分布。第四是你训练多长时间。第五是你需要一个能无限扩展的目标函数。预训练的目标函数就是这样的一个目标函数。另一个是 RL 目标函数，它说你有一个目标，你要去达成这个目标。在这个框架内，有像数学和编程中的客观奖励，也有像 RLHF 或其更高级版本中的主观奖励。第六和第七是关于归一化或调节的事情，就是确保数值稳定性，让这团巨大的计算以层流方式运行，而不是遇到各种问题。这就是那个假说，也是我至今仍然持有的假说。我没有看到太多与它不一致的东西。预训练的 scaling laws 是其中一个例子。那些一直在继续。现在已经被广泛报道了，我们对预训练感觉很好。它持续带来收益。变化在于，现在我们也在 RL 上看到了同样的事情。我们看到先有一个预训练阶段，然后在其之上有一个 RL 阶段。对于 RL，实际上完全一样。甚至其他公司在一些发布中也公布过类似的东西，说"我们在数学竞赛——AIME 或其他比赛——上训练模型，模型的表现与训练时长呈对数线性关系。"我们也看到了这一点，而且不仅仅是数学竞赛。这是在广泛的 RL 任务上的。我们在 RL 上看到了和预训练时一样的 scaling。
**Dario Amodei:** I actually have the same hypothesis I had even all the way back in 2017. I think I talked about it last time, but I wrote a doc called "The Big Blob of Compute Hypothesis". It wasn't about the scaling of language models in particular. When I wrote it GPT-1 had just come out. That was one among many things. Back in those days there was robotics. People tried to work on reasoning as a separate thing from language models, and there was scaling of the kind of RL that happened in AlphaGo and in Dota at OpenAI. People remember StarCraft at DeepMind, AlphaStar. It was written as a more general document. Rich Sutton put out "The Bitter Lesson" a couple years later. The hypothesis is basically the same. What it says is that all the cleverness, all the techniques, all the "we need a new method to do something", that doesn't matter very much. There are only a few things that matter. I think I listed seven of them. One is how much raw compute you have. The second is the quantity of data. The third is the quality and distribution of data. It needs to be a broad distribution. The fourth is how long you train for. The fifth is that you need an objective function that can scale to the moon. The pre-training objective function is one such objective function. Another is the RL objective function that says you have a goal, you're going to go out and reach the goal. Within that, there's objective rewards like you see in math and coding, and there's more subjective rewards like you see in RLHF or higher-order versions of that. Then the sixth and seventh were things around normalization or conditioning, just getting the numerical stability so that the big blob of compute flows in this laminar way instead of running into problems. That was the hypothesis, and it's a hypothesis I still hold. I don't think I've seen very much that is not in line with it. The pre-training scaling laws were one example of what we see there. Those have continued going. Now it's been widely reported, we feel good about pre-training. It's continuing to give us gains. What has changed is that now we're also seeing the same thing for RL. We're seeing a pre-training phase and then an RL phase on top of that. With RL, it's actually just the same. Even other companies have published things in some of their releases that say, "We train the model on math contests — AIME or other things — and how well the model does is log-linear in how long we've trained it." We see that as well, and it's not just math contests. It's a wide variety of RL tasks. We're seeing the same scaling in RL that we saw for pre-training.
**Dwarkesh Patel：** 你提到了 Rich Sutton 和"The Bitter Lesson"。我去年采访过他，他其实非常不认同 LLM 的路线。我不确定这是否完全是他的观点，但可以这样概括他的反对意见：一个真正拥有人类学习核心能力的东西，不应该需要这些数十亿美元的数据和算力，以及这些定制化的环境，来学会使用 Excel、PowerPoint、浏览网页。我们必须用这些 RL 环境来内嵌这些技能，这一事实暗示我们实际上缺少一个核心的人类学习算法。所以我们在 scaling 错误的东西。这确实引出了一个问题。如果我们认为会有某种像人类一样能即时学习的东西，为什么我们还在做所有这些 RL scaling？
**Dwarkesh Patel:** You mentioned Rich Sutton and "The Bitter Lesson". I interviewed him last year, and he's actually very non-LLM-pilled. I don't know if this is his perspective, but one way to paraphrase his objection is: Something which possesses the true core of human learning would not require all these billions of dollars of data and compute and these bespoke environments, to learn how to use Excel, how to use PowerPoint, how to navigate a web browser. The fact that we have to build in these skills using these RL environments hints that we are actually lacking a core human learning algorithm. So we're scaling the wrong thing. That does raise the question. Why are we doing all this RL scaling if we think there's something that's going to be human-like in its ability to learn on the fly?
**Dario Amodei：** 我认为这把几个应该分开思考的事情混在了一起。这里确实有一个真正的谜题，但它可能并不重要。事实上，我猜它可能不重要。有一个有趣的现象。让我先把 RL 放在一边，因为我实际上认为把 RL 和预训练区别对待是一个误导。如果我们看预训练 scaling，2017 年 Alec Radford 做 GPT-1 的时候就很有趣。GPT-1 之前的模型是在不代表广泛文本分布的数据集上训练的。有非常标准的语言建模基准。GPT-1 本身是在一堆同人小说上训练的，我记得是的。那是文学文本，只是你能获取的文本中非常小的一部分。那时候大概是十亿词左右，很小的数据集，代表着你能看到的世界的一个很窄的分布。它的泛化能力不好。如果你在某个同人小说语料库上做得更好，并不能很好地泛化到其他任务。我们有各种各样的衡量标准。我们有各种衡量它在预测其他类型文本上表现如何的标准。只有当你在互联网上的所有任务上训练——当你做一个通用的互联网爬取，比如 Common Crawl 或者爬取 Reddit 上的链接，这是我们在 GPT-2 时做的——你才开始获得泛化能力。我认为我们在 RL 上看到了同样的事情。我们先从简单的 RL 任务开始，比如在数学竞赛上训练，然后转向更广泛的训练，包括代码之类的。现在我们正在转向许多其他任务。我认为然后我们会越来越获得泛化能力。所以这算是化解了 RL 与预训练之间的对立。但无论如何都有一个谜题，就是在预训练中我们使用了万亿级的 token。人类看不到万亿级的词。所以这里确实存在样本效率的差异。模型从零开始，需要更多训练。但我们也看到，一旦训练完成，如果我们给它们一百万的长上下文——唯一阻碍长上下文的是推理——它们在那个上下文中学习和适应的能力非常强。所以我不完全知道答案。我认为有什么正在发生，预训练不像人类学习的过程，但它介于人类学习和人类进化之间。我们的很多先验知识来自进化。我们的大脑不是白板。关于这个已经写了很多书。语言模型更像白板。它们真的从随机权重开始，而人脑起步时就有所有这些区域连接到各种输入和输出。也许我们应该把预训练——以及 RL——看作存在于人类进化和人类即时学习之间的中间地带。我们应该把模型的上下文学习看作介于人类长期学习和短期学习之间的东西。所以有这样一个层次结构。有进化，有长期学习，有短期学习，还有人类的即时反应。LLM 的各个阶段存在于这个光谱上，但不一定在完全相同的点上。模型没有和某些人类学习模式的类比物，它们落在这些点之间。这说得通吗？
**Dario Amodei:** I think this puts together several things that should be thought of differently. There is a genuine puzzle here, but it may not matter. In fact, I would guess it probably doesn't matter. There is an interesting thing. Let me take the RL out of it for a second, because I actually think it's a red herring to say that RL is any different from pre-training in this matter. If we look at pre-training scaling, it was very interesting back in 2017 when Alec Radford was doing GPT-1. The models before GPT-1 were trained on datasets that didn't represent a wide distribution of text. You had very standard language modeling benchmarks. GPT-1 itself was trained on a bunch of fanfiction, I think actually. It was literary text, which is a very small fraction of the text you can get. In those days it was like a billion words or something, so small datasets representing a pretty narrow distribution of what you can see in the world. It didn't generalize well. If you did better on some fanfiction corpus, it wouldn't generalize that well to other tasks. We had all these measures. We had all these measures of how well it did at predicting all these other kinds of texts. It was only when you trained over all the tasks on the internet — when you did a general internet scrape from something like Common Crawl or scraping links in Reddit, which is what we did for GPT-2 — that you started to get generalization. I think we're seeing the same thing on RL. We're starting first with simple RL tasks like training on math competitions, then moving to broader training that involves things like code. Now we're moving to many other tasks. I think then we're going to increasingly get generalization. So that kind of takes out the RL vs. pre-training side of it.
**Dwarkesh Patel：** 是的，虽然有些地方还是有点让人困惑。比如，如果类比是说这就像进化，所以样本效率低是正常的，那么如果我们要从上下文学习中获得超级样本高效的 agent，为什么我们还要费力去构建所有这些 RL 环境？有些公司的工作似乎就是教模型如何使用某个 API、如何用 Slack、如何用各种工具。如果那种能即时学习的 agent 正在涌现或者已经涌现了，为什么还要在这上面花这么大力气，这让我很困惑。
**Dwarkesh Patel:** But there is a puzzle either way, which is that in pre-training we use trillions of tokens. Humans don't see trillions of words. So there is an actual sample efficiency difference here. There is actually something different here. The models start from scratch and they need much more training. But we also see that once they're trained, if we give them a long context length of a million — the only thing blocking long context is inference — they're very good at learning and adapting within that context. So I don't know the full answer to this. I think there's something going on where pre-training is not like the process of humans learning, but it's somewhere between the process of humans learning and the process of human evolution. We get many of our priors from evolution. Our brain isn't just a blank slate. Whole books have been written about this. The language models are much more like blank slates. They literally start as random weights, whereas the human brain starts with all these regions connected to all these inputs and outputs. Maybe we should think of pre-training — and for that matter, RL as well — as something that exists in the middle space between human evolution and human on-the-spot learning. And we should think of the in-context learning that the models do as something between long-term human learning and short-term human learning. So there's this hierarchy. There's evolution, there's long-term learning, there's short-term learning, and there's just human reaction. The LLM phases exist along this spectrum, but not necessarily at exactly the same points. There's no analog to some of the human modes of learning the LLMs are falling in between the points. Does that make sense?
**Dario Amodei：** 我没法代表其他人的重点。我只能谈我们是怎么想的。目标不是在 RL 中教模型每一个可能的技能，就像我们在预训练中也不是这样做的。在预训练中，我们不是试图让模型接触每一种可能的文字组合方式。相反，模型在很多东西上训练，然后在预训练中达到泛化。那就是从 GPT-1 到 GPT-2 的转变，我亲眼见证了。模型达到了一个临界点。我有过那种时刻，"哦对，你只要给模型一列数字——这是房子的价格，这是房子的面积——模型就能补全模式，做线性回归。"不太精确，但它能做到，而且它从来没有见过那个确切的东西。所以，我们在构建这些 RL 环境的程度上，目标和五到十年前在预训练中做的非常相似。我们试图获取大量数据，不是因为我们想覆盖某个特定文档或特定技能，而是因为我们想要泛化。
**Dario Amodei:** Yes, although some things are still a bit confusing. For example, if the analogy is that this is like evolution so it's fine that it's not sample efficient, then if we're going to get super sample-efficient agent from in-context learning, why are we bothering to build all these RL environments? There are companies whose work seems to be teaching models how to use this API, how to use Slack, how to use whatever. It's confusing to me why there's so much emphasis on that if the kind of agent that can just learn on the fly is emerging or has already emerged.
**Dwarkesh Patel：** 我觉得你搭建的这个框架显然是有道理的。我们在朝 AGI 迈进。现在没人会否认我们在本世纪内会实现 AGI。关键分歧在于，你说我们正在接近指数增长的终点。而另一些人看着这个说，"我们从 2012 年以来一直在进步，到 2035 年我们会有类人的 agent。"显然我们在这些模型中看到了进化做过的事，或者人的一生学习中做过的事。我想了解你看到了什么，让你觉得是一年后而不是十年后。
**Dwarkesh Patel:** I can't speak for the emphasis of anyone else. I can only talk about how we think about it. The goal is not to teach the model every possible skill within RL, just as we don't do that within pre-training. Within pre-training, we're not trying to expose the model to every possible way that words could be put together. Rather, the model trains on a lot of things and then reaches generalization across pre-training. That was the transition from GPT-1 to GPT-2 that I saw up close. The model reaches a point. I had these moments where I was like, "Oh yeah, you just give the model a list of numbers — this is the cost of the house, this is the square feet of the house — and the model completes the pattern and does linear regression." Not great, but it does it, and it's never seen that exact thing before. So to the extent that we are building these RL environments, the goal is very similar to what was done five or ten years ago with pre-training. We're trying to get a whole bunch of data, not because we want to cover a specific document or a specific skill, but because we want to generalize.
**Dario Amodei：** 这里可以做两种声明，一个更强一个更弱。从更弱的声明开始，当我在 2019 年第一次看到 scaling 的时候，我并不确定。这是一个 50/50 的事情。我以为我看到了什么。我的声明是这件事比任何人认为的都更有可能。也许有 50% 的概率会发生。对于基本假设——用你的话说，十年内我们会达到我所说的"数据中心中的天才之国"——我现在有 90% 的信心。很难超过 90%，因为世界太不可预测了。也许不可消除的不确定性让我们到 95%，那就涉及到多家公司内部动荡、台湾被入侵、所有晶圆厂被导弹炸毁之类的情况。
**Dario Amodei:** I think the framework you're laying down obviously makes sense. We're making progress toward AGI. Nobody at this point disagrees we're going to achieve AGI this century. The crux is you say we're hitting the end of the exponential. Somebody else looks at this and says, "We've been making progress since 2012, and by 2035 we'll have a human-like agent." Obviously we're seeing in these models the kinds of things that evolution did, or that learning within a human lifetime does. I want to understand what you're seeing that makes you think it's one year away and not ten years away.
**Dwarkesh Patel：** 你这是在给我们招祸，Dario。
**Dwarkesh Patel:** There are two claims you could make here, one stronger and one weaker. Starting with the weaker claim, when I first saw the scaling back in 2019, I wasn't sure. This was a 50/50 thing. I thought I saw something. My claim was that this was much more likely than anyone thinks. Maybe there's a 50% chance this happens. On the basic hypothesis of, as you put it, within ten years we'll get to what I call a "country of geniuses in a data center", I'm at 90% on that. It's hard to go much higher than 90% because the world is so unpredictable. Maybe the irreducible uncertainty puts us at 95%, where you get to things like multiple companies having internal turmoil, Taiwan gets invaded, all the fabs get blown up by missiles.
**Dario Amodei：** 你可以构建一个 5% 的世界，事情被延迟十年。还有另外 5% 是——我对可验证的任务非常有信心。对于编程，除了那些不可消除的不确定性，我认为一两年内我们就会到达那里。十年内我们在端到端编程上不可能到不了那里。我唯一一点根本性的不确定，即使在长时间尺度上，是关于那些不可验证的任务：规划火星任务、做像 CRISPR 这样的基础科学发现、写小说。这些任务很难验证。我几乎确定我们有一条可靠的路径到达那里，但如果有一点不确定性的话，就在那里。十年的时间线上我有 90% 的信心，这几乎是你能达到的最高确定度了。我觉得说 2035 年之前这不会发生是疯狂的。在一个理性的世界里，这应该是主流共识之外的观点。
**Dario Amodei:** Now you've jinxed us, Dario.
**Dwarkesh Patel：** 但对验证的强调在我看来暗示着对这些模型泛化能力的信心不足。如果你想想人类，我们既擅长有可验证奖励的事情，也擅长没有可验证奖励的事情。
**Dwarkesh Patel:** You could construct a 5% world where things get delayed for ten years. There's another 5% which is that I'm very confident on tasks that can be verified. With coding, except for that irreducible uncertainty, I think we'll be there in one or two years. There's no way we will not be there in ten years in terms of being able to do end-to-end coding. My one little bit of fundamental uncertainty, even on long timescales, is about tasks that aren't verifiable: planning a mission to Mars; doing some fundamental scientific discovery like CRISPR; writing a novel. It's hard to verify those tasks. I am almost certain we have a reliable path to get there, but if there's a little bit of uncertainty it's there. On the ten-year timeline I'm at 90%, which is about as certain as you can be. I think it's crazy to say that this won't happen by 2035. In some sane world, it would be outside the mainstream.
**Dario Amodei：** 不，这正是为什么我几乎是确定的。我们已经看到了从可验证的事物到不可验证的事物之间的大量泛化。我们已经在看到这个了。
**Dario Amodei:** But the emphasis on verification hints to me a lack of belief that these models are generalized. If you think about humans, we're both good at things for which we get verifiable reward and things for which we don't.
**Dwarkesh Patel：** 但你似乎在强调这是一个光谱，会把不同领域的进展分开。这看起来不像人类变好的方式。
**Dwarkesh Patel:** No, this is why I'm almost sure. We already see substantial generalization from things that verify to things that don't. We're already seeing that.
**Dario Amodei：** 我们到不了那里的世界是这样的：我们做完所有可验证的事情。很多会泛化，但我们没有完全到达。我们没有完全填满另一边。这不是一个二元的事情。即使泛化很弱，你只能做可验证的领域，我也不确定在这样一个世界里你能自动化软件工程。你在某种意义上是"一个软件工程师"，但作为软件工程师的一部分，对你来说包括写关于你宏大愿景的长篇备忘录。
**Dario Amodei:** But it seems like you were emphasizing this as a spectrum which will split apart which domains in which we see more progress. That doesn't seem like how humans get better.
**Dwarkesh Patel：** 我不认为那是 SWE 工作的一部分。那是公司的工作，不是 SWE 特有的。
**Dwarkesh Patel:** The world in which we don't get there is the world in which we do all the verifiable things. Many of them generalize, but we don't fully get there. We don't fully color in the other side of the box. It's not a binary thing. Even if generalization is weak and you can only do verifiable domains, it's not clear to me you could automate software engineering in such a world. You are "a software engineer" in some sense, but part of being a software engineer for you involves writing long memos about your grand vision.
**Dario Amodei：** 但 SWE 确实涉及设计文档和其他类似的东西。模型已经很擅长写注释了。再说一遍，我在这里做的声明比我实际相信的要弱得多，以区分两件事。我们在软件工程上几乎已经到了。
**Dario Amodei:** I don't think that's part of the job of SWE. That's part of the job of the company, not SWE specifically.
**Dwarkesh Patel：** 用什么指标衡量？有一个指标是 AI 写了多少行代码。如果你考虑软件工程历史上的其他生产力提升，编译器写了所有的软件代码行。写了多少行和生产力提升有多大之间是有区别的。"我们几乎到了"是指……生产力提升有多大，而不仅仅是 AI 写了多少行？
**Dwarkesh Patel:** But SWE does involve design documents and other things like that. The models are already pretty good at writing comments. Again, I'm making much weaker claims here than I believe, to distinguish between two things. We're already almost there for software engineering.
**Dario Amodei：** 我实际上同意你说的。我对代码和软件工程做了一系列预测。我觉得人们反复误解了它们。让我把这个光谱铺开。大约八九个月前，我说 AI 模型将在三到六个月内写出 90% 的代码行。这实现了，至少在某些地方。在 Anthropic 实现了，在很多使用我们模型的下游用户那里也实现了。但这实际上是一个非常弱的标准。人们以为我在说我们不需要 90% 的软件工程师了。这两件事天差地别。这个光谱是：模型写 90% 的代码，模型写 100% 的代码。这在生产力上是一个巨大的差异。90% 的端到端 SWE 任务——包括编译、设置集群和环境、测试功能、写备忘录等——由模型完成。100% 的当前 SWE 任务由模型完成。即使这发生了，也不意味着软件工程师失业了。他们可以做新的、更高层次的事情，他们可以管理。然后再往光谱的更远处，是对 SWE 的需求减少 90%，我认为这会发生，但这是一个光谱。我在"The Adolescence of Technology"（技术的青春期）中写过，用农业的例子讨论了这种光谱。
**Dario Amodei:** By what metric? There's one metric which is how many lines of code are written by AI. If you consider other productivity improvements in the history of software engineering, compilers write all the lines of software. There's a difference between how many lines are written and how big the productivity improvement is. "We're almost there" meaning… How big is the productivity improvement, not just how many lines are written by AI?
**Dwarkesh Patel：** 我完全同意你说的。这些是非常不同的基准，但我们正在超快速地穿越它们。你的愿景的一部分是从 90 到 100 会很快发生，而且会带来巨大的生产力提升。但我注意到的是，即使在全新项目中，人们一开始就用 Claude Code 之类的工具，人们报告说他们启动了很多项目……我们在外面的世界看到软件的复兴了吗？看到了所有这些本来不会存在的新功能了吗？至少到目前为止，似乎我们没有看到。所以这确实让我怀疑。即使我从来不需要干预 Claude Code，世界是复杂的。工作是复杂的。在自包含系统上闭环，无论是写软件还是其他什么，我们到底能看到多大的更广泛收益？也许这应该稀释我们对"天才之国"的估计。
**Dwarkesh Patel:** I actually agree with you on this. I've made a series of predictions on code and software engineering. I think people have repeatedly misunderstood them. Let me lay out the spectrum. About eight or nine months ago, I said the AI model will be writing 90% of the lines of code in three to six months. That happened, at least at some places. It happened at Anthropic, happened with many people downstream using our models. But that's actually a very weak criterion. People thought I was saying that we won't need 90% of the software engineers. Those things are worlds apart. The spectrum is: 90% of code is written by the model, 100% of code is written by the model. That's a big difference in productivity. 90% of the end-to-end SWE tasks — including things like compiling, setting up clusters and environments, testing features, writing memos — are done by the models. 100% of today's SWE tasks are done by the models. Even when that happens, it doesn't mean software engineers are out of a job. There are new higher-level things they can do, where they can manage. Then further down the spectrum, there's 90% less demand for SWEs, which I think will happen but this is a spectrum. I wrote about it in "The Adolescence of Technology" where I went through this kind of spectrum with farming.
**Dario Amodei：** 我同时同意你说的，这是这些事情不会瞬间发生的一个原因，但同时，我认为效果会非常快。你可以有两个极端。一个是 AI 不会取得进展。它很慢。在经济中扩散需要很长时间。经济扩散已经变成了那种用来说明 AI 不会取得进展或 AI 进展不重要的流行词。另一个极端是我们将获得递归自我改进，一切都会加速。你能在曲线上画一条指数线吗？我们会在获得递归后的多少纳秒内就拥有环绕太阳的戴森球。我在这里完全在夸张这个观点，但确实存在这两个极端。但从一开始我们看到的，至少在 Anthropic 内部看到的，是这种奇特的每年 10 倍的收入增长。2023 年是从零到 1 亿美元。2024 年是从 1 亿到 10 亿美元。2025 年是从 10 亿到 90-100 亿美元。
**Dario Amodei:** I actually totally agree with you on that. These are very different benchmarks from each other, but we're proceeding through them super fast. Part of your vision is that going from 90 to 100 is going to happen fast, and that it leads to huge productivity improvements. But what I notice is that even in greenfield projects people start with Claude Code or something, people report starting a lot of projects… Do we see in the world out there a renaissance of software, all these new features that wouldn't exist otherwise? At least so far, it doesn't seem like we see that. So that does make me wonder. Even if I never had to intervene with Claude Code, the world is complicated. Jobs are complicated. Closing the loop on self-contained systems, whether it's just writing software or something, how much broader gains would we see just from that? Maybe that should dilute our estimation of the "country of geniuses".
**Dwarkesh Patel：** 你们应该买 10 亿美元自己的产品，这样就能……
**Dwarkesh Patel:** I simultaneously agree with you that it's a reason why these things don't happen instantly, but at the same time, I think the effect is gonna be very fast. You could have these two poles. One is that AI is not going to make progress. It's slow. It's going to take forever to diffuse within the economy. Economic diffusion has become one of these buzzwords that's a reason why we're not going to make AI progress, or why AI progress doesn't matter. The other axis is that we'll get recursive self-improvement, the whole thing. Can't you just draw an exponential line on the curve? We're going to have Dyson spheres around the sun so many nanoseconds after we get recursive. I'm completely caricaturing the view here, but there are these two extremes. But what we've seen from the beginning, at least if you look within Anthropic, there's this bizarre 10x per year growth in revenue that we've seen. So in 2023, it was zero to $100 million. In 2024, it was $100 million to $1 billion. In 2025, it was $1 billion to $ 9-10 billion.
**Dario Amodei：** 而且今年第一个月，那条指数曲线……你会以为它会放缓，但我们在一月份就又增加了几十亿的收入。显然这条曲线不可能永远持续下去。GDP 就那么大。我甚至猜测今年会有所弯曲，但那是一条很快的曲线。那真的是一条非常快的曲线。我打赌即使规模扩展到整个经济体，它仍然会保持相当快。所以我认为我们应该思考的是这个中间世界，事情极其迅速，但不是瞬间的，它们需要时间，因为经济扩散，因为需要闭环。因为很琐碎："我必须在企业内部做变更管理……我设置好了这个，但我必须更改安全权限才能让它真正运行……我有一个旧软件在编译和发布前会检查模型，我得重写它。是的，模型能做到，但我得告诉模型去做。它需要时间来做这些事。"所以我认为到目前为止我们看到的一切都与这个想法一致：有一条快速的指数曲线是模型的能力。然后有另一条快速的指数曲线是下游的，就是模型向经济中的扩散。不是瞬间的，不是慢的，比以前任何技术都快得多，但有其极限。当我看 Anthropic 内部，当我看我们的客户：快速采用，但不是无限快。
**Dario Amodei:** You guys should have just bought a billion dollars of your own products so you could just… And the first month of this year, that exponential is... You would think it would slow down, but we added another few billion to revenue in January. Obviously that curve can't go on forever. The GDP is only so large. I would even guess that it bends somewhat this year, but that is a fast curve. That's a really fast curve. I would bet it stays pretty fast even as the scale goes to the entire economy. So I think we should be thinking about this middle world where things are extremely fast, but not instant, where they take time because of economic diffusion, because of the need to close the loop. Because it's fiddly: "I have to do change management within my enterprise… I set this up, but I have to change the security permissions on this in order to make it actually work… I had this old piece of software that checks the model before it's compiled and released and I have to rewrite it. Yes, the model can do that, but I have to tell the model to do that. It has to take time to do that." So I think everything we've seen so far is compatible with the idea that there's one fast exponential that's the capability of the model. Then there's another fast exponential that's downstream of that, which is the diffusion of the model into the economy.
**Dwarkesh Patel：** 我能试一个犀利的观点吗？
**Dwarkesh Patel:** Not instant, not slow, much faster than any previous technology, but it has its limits. When I look inside Anthropic, when I look at our customers: fast adoption, but not infinitely fast. Can I try a hot take on you? Yeah. I feel like diffusion is cope that people say. When the model isn't able to do something, they're like, "oh, but it's a diffusion issue." But then you should use the comparison to humans. You would think that the inherent advantages that AIs have would make diffusion a much easier problem for new AIs getting onboarded than new humans getting onboarded. An AI can read your entire Slack and your drive in minutes. They can share all the knowledge that the other copies of the same instance have. You don't have this adverse selection problem when you're hiring AI, so you can just hire copies of a vetted AI model. Hiring a human is so much more of a hassle. People hire humans all the time. We pay humans upwards of $50 trillion in wages because they're useful, even though in principle it would be much easier to integrate AIs into the economy than it is to hire humans. The diffusion doesn't really explain. I think diffusion
**Dario Amodei：** 好啊。
**Dario Amodei:** wages because they're useful, even though in principle it would be much easier to integrate AIs into the economy than it is to hire humans. The diffusion doesn't really explain.
**Dwarkesh Patel：** 我觉得扩散论是人们的自我安慰。当模型做不到某件事时，他们就说"哦，但这是扩散问题。"但你应该和人类做比较。你会觉得 AI 的内在优势会让扩散问题比新人类入职容易得多。AI 可以在几分钟内读完你所有的 Slack 和 Drive。它们可以共享同一个实例的所有其他副本拥有的知识。你在雇佣 AI 时没有逆向选择问题，所以你可以直接雇佣经过验证的 AI 模型的副本。雇佣人类要麻烦得多。人们一直在雇佣人类。我们每年支付超过 50 万亿美元的人类工资，因为他们有用，尽管原则上把 AI 整合到经济中比雇佣人类容易得多。扩散并不能真正解释这个。
**Dwarkesh Patel:** I think diffusion is very real and doesn't exclusively have to do with limitations on the AI models. Again, there are people who use diffusion as kind of a buzzword to say this isn't a big deal. I'm not talking about that. I'm not talking about how AI will diffuse at the speed of previous technologies. I think AI will diffuse much faster than previous technologies have, but not infinitely fast. I'll just give an example of this. There's Claude Code. Claude Code is extremely easy to set up. If you're a developer, you can just start using Claude Code. There is no reason why a developer at a large enterprise should not be adopting Claude Code as quickly as an individual developer or developer at a startup. We do everything we can to promote it. We sell Claude Code to enterprises. Big enterprises, big financial companies, big pharmaceutical companies, all of them are adopting Claude Code much faster than enterprises typically adopt new technology. But again, it takes time. Any given feature or any given product, like Claude Code or Cowork, will get adopted by the individual developers who are on Twitter all the time, by the Series A startups, many months faster than they will get adopted by a large enterprise that does food sales. There are just a number of factors. You have to go through legal, you have to provision it for everyone. It has to pass security and compliance. The leaders of the company who are further away from the AI revolution are forward-looking, but they have to say, "Oh, it makes sense for us to spend 50 million. This is what this Claude Code thing is. This is why it helps our company. This is why it makes us more productive." Then they have to explain to the people two levels below. They have to say, "Okay, we have 3,000 developers. Here's how we're going to roll it out to our developers." We have conversations like this every day. We are doing everything we can to make Anthropic's revenue grow 20 or 30x a year instead of 10x a year. Again, many enterprises are just saying, "This is so productive. We're going to take shortcuts in our usual procurement process." They're moving much faster than when we tried to sell them just the ordinary API, which many of them use. Claude Code is a more compelling product, but it's not an infinitely compelling product. I don't think even AGI or powerful AI or "country of geniuses in a data center" will be an infinitely compelling product. It will be a compelling product enough maybe to get 3-5x, or 10x, a year of growth, even when you're in the hundreds of billions of dollars, which is extremely hard to do and has never been done in history before, but not infinitely fast.
**Dario Amodei：** 我认为扩散是非常真实的，而且不完全与 AI 模型的局限性有关。再说一遍，有些人把扩散当作一个流行词来说这没什么大不了的。我不是在说那个。我不是在说 AI 会以以前技术的速度扩散。我认为 AI 的扩散速度会比以前的技术快得多，但不是无限快。我就举一个例子。有 Claude Code。Claude Code 非常容易设置。如果你是开发者，你可以直接开始用 Claude Code。大型企业的开发者没有理由不像个人开发者或初创公司的开发者一样快速采用 Claude Code。我们尽一切努力推广它。我们向企业销售 Claude Code。大型企业、大型金融公司、大型制药公司，它们都在以比企业通常采用新技术快得多的速度采用 Claude Code。但再说一遍，这需要时间。任何特定功能或产品，比如 Claude Code 或 Cowork，会被那些整天泡在 Twitter 上的个人开发者、A 轮初创公司，比被一家做食品销售的大型企业采用要早好几个月。就是有很多因素。你得过法务审批，你得为每个人配置。它必须通过安全和合规审查。公司领导离 AI 革命更远一些，他们是有前瞻性的，但他们得说，"好的，我们花 5000 万美元有道理。这就是 Claude Code 这个东西。这就是它为什么能帮助我们公司。这就是它为什么让我们更高效。"然后他们得向下面两级的人解释。他们得说，"好的，我们有 3000 个开发者。这是我们打算怎样向开发者推出它。"我们每天都有这样的对话。我们尽一切努力让 Anthropic 的收入每年增长 20 或 30 倍，而不是 10 倍。再说一遍，很多企业就是在说，"这太高效了。我们要在常规采购流程上抄近路。"他们比我们试图卖给他们普通 API 时快得多。Claude Code 是一个更有吸引力的产品，但它不是一个无限有吸引力的产品。我不认为即使是 AGI 或强大 AI 或"数据中心中的天才之国"会是一个无限有吸引力的产品。它会足够有吸引力，也许能实现每年 3-5 倍或 10 倍的增长，即使在数千亿美元的规模上，这极其难做到，在历史上从未有过，但不是无限快。
**Dario Amodei:** I buy that it would be a slight slowdown. Maybe this is not your claim, but sometimes people talk about this like, "Oh, the capabilities are there, but because of diffusion... otherwise we're basically at AGI". I don't believe we're basically at AGI. I think if you had the "country of geniuses in a data center"...
**Dwarkesh Patel：** 我接受会有一些减速。也许这不是你的声明，但有时人们这样说，"哦，能力已经在那里了，但因为扩散……否则我们基本上已经是 AGI 了"。我不相信我们基本上已经是 AGI 了。我认为如果你有"数据中心中的天才之国"……
**Dwarkesh Patel:** If we had the "country of geniuses in a data center", we would know it. We would know it if you had the "country of geniuses in a data center". Everyone in this room would know it. Everyone in Washington would know it. People in rural parts might not know it, but we would know it. We don't have that now. That is very clear.
**Dario Amodei：** 如果我们有"数据中心中的天才之国"，我们会知道的。如果你有"数据中心中的天才之国"，我们会知道的。这个房间里的每个人都会知道。华盛顿的每个人都会知道。农村地区的人可能不知道，但我们会知道。我们现在没有那个。这是非常明确的。
**Dario Amodei:** Coming back to concrete prediction… Because there are so many different things to disambiguate, it can be easy to talk past each other when we're talking about capabilities. For example, when I interviewed you three years ago, I asked you a prediction about what we should expect three years from now. You were right. You said, "We should expect systems which, if you talk to them for the course of an hour, it's hard to tell them apart from a generally well-educated human." I think you were right about that. I think spiritually I feel unsatisfied because my internal expectation was that such a system could automate large parts of white-collar work. So it might be more productive to talk about the actual end capabilities you want from such a system.
**Dwarkesh Patel：** 回到具体预测……因为有太多不同的东西需要区分，当我们谈论能力时很容易各说各话。比如，三年前我采访你时，我问你对三年后应该期待什么的预测。你说对了。你说，"我们应该期待这样的系统：如果你和它们对话一个小时，很难把它们和一个受过良好教育的普通人区分开来。"我觉得你说对了。但在精神上我觉得不满足，因为我内心的期待是这样的系统应该能自动化大部分白领工作。所以谈论你想要的实际最终能力可能更有成效。
**Dwarkesh Patel:** I will basically tell you where I think we are.
**Dario Amodei：** 我基本上会告诉你我认为我们现在在哪里。
**Dario Amodei:** Let me ask a very specific question so that we can figure out exactly what kinds of capabilities we should think about soon. Maybe I'll ask about it in the context of a job I understand well, not because it's the most relevant job, but just because I can evaluate the claims about it. Take video editors. I have video editors. Part of their job involves learning about our audience's preferences, learning about my preferences and tastes, and the different trade-offs we have. They're, over the course of many months, building up this understanding of context. The skill and ability they have six months into the job, a model that can pick up that skill on the job on the fly, when should we expect such an AI system?
**Dwarkesh Patel：** 让我问一个非常具体的问题，这样我们就能准确弄清楚我们应该思考什么样的近期能力。也许我用一个我了解的工作来问，不是因为它最相关，而是因为我能评估关于它的声明。拿视频编辑来说。我有视频编辑。他们工作的一部分涉及了解我们受众的偏好，了解我的偏好和品味，以及我们面对的各种取舍。他们在很多个月里逐渐建立对上下文的理解。他们入职六个月后拥有的技能和能力——一个能在工作中即时习得那种技能的模型，我们什么时候应该期待这样的 AI 系统？
**Dwarkesh Patel:** I guess what you're talking about is that we're doing this interview for three hours. Someone's going to come in, someone's going to edit it. They're going to be like, "Oh, I don't know, Dario scratched his head and we could edit that out." "Magnify that." "There was this long discussion that is less interesting to people. There's another thing that's more interesting to people, so let's make this edit." I think the "country of geniuses in a data center" will be able to do that. The way it will be able to do that is it will have general control of a computer screen. You'll be able to feed this in. It'll be able to also use the computer screen to go on the web, look at all your previous interviews, look at what people are saying on Twitter in response to your interviews, talk to you, ask you questions, talk to your staff, look at the history of edits that you did, and from that, do the job. I think that's dependent on several things. I think this is one of the things that's actually blocking deployment: getting to the point on computer use where the models are really masters at using the computer. We've seen this climb in benchmarks, and benchmarks are always imperfect measures. But I think when we first released computer use a year and a quarter ago, OSWorld was at maybe 15%. I don't remember exactly, but we've climbed from that to 65-70%. There may be harder measures as well, but I think computer use has to pass a point of reliability.
**Dario Amodei：** 我猜你说的是，我们在做这次三小时的访谈。会有人来剪辑。他们会说，"哦，我不知道，Dario 挠了挠头，我们可以把那个剪掉。""放大那个。""有一段较长的讨论对观众来说不那么有趣。另一个内容对观众来说更有趣，所以我们做这个剪辑。"我认为"数据中心中的天才之国"将能做到这些。它能做到的方式是，它将对电脑屏幕有通用控制能力。你可以把视频输入进去。它还能使用电脑屏幕上网，查看你之前所有的访谈，查看人们在 Twitter 上对你访谈的评论，和你交谈，问你问题，和你的团队交谈，查看你做过的剪辑历史，然后从中完成工作。我认为这取决于几件事。我认为这是实际上阻碍部署的事情之一：让模型在电脑操作上达到真正精通的水平。我们看到了基准测试的攀升，基准测试永远是不完美的衡量标准。但我认为当我们一年零一个季度前首次发布 computer use 时，OSWorld 大约在 15%。我不记得确切数字，但我们从那里攀升到了 65-70%。可能还有更难的测量标准，但我认为 computer use 必须通过一个可靠性的临界点。
**Dario Amodei:** Can I just follow up on that before you move on to the next point? For years, I've been trying to build different internal LLM tools for myself. Often I have these text-in, text-out tasks, which should be dead center in the repertoire of these models. Yet I still hire humans to do them. If it's something like, "identify what the best clips would be in this transcript", maybe the LLMs do a seven-out-of-ten job on them. But there's not this ongoing way I can engage with them to help them get better at the job the way I could with a human employee. That missing ability, even if you solve computer use, would still block my ability to offload an actual job to them.
**Dwarkesh Patel：** 我能在你继续之前追问一下吗？多年来，我一直试图为自己构建各种内部 LLM 工具。通常我有一些文本输入、文本输出的任务，这些应该完全在这些模型的能力范围之内。但我仍然雇佣人类来做。如果是"识别这份转录稿中最好的片段"，也许 LLM 能做到七分的水平。但没有一种持续的方式让我和它们互动来帮助它们把工作做得更好，就像我能和人类员工做的那样。这种缺失的能力，即使你解决了 computer use，仍然会阻碍我把一份实际工作交给它们。
**Dwarkesh Patel:** This gets back to what we were talking about before with learning on the job. It's very interesting. I think with the coding agents, I don't think people would say that learning on the job is what is preventing the coding agents from doing everything end to end. They keep getting better. We have engineers at Anthropic who don't write any code. When I look at the productivity, to your previous question, we have folks who say, "This GPU kernel, this chip, I used to write it myself. I just have Claude do it." There's this enormous improvement in productivity. When I see Claude Code, familiarity with the codebase or a feeling that the model hasn't worked at the company for a year, that's not high up on the list of complaints I see. I think what I'm saying is that we're kind of taking a different path.
**Dario Amodei：** 这回到了我们之前讨论的在职学习的问题。这很有趣。我认为对于编程 agent，人们不会说在职学习是阻碍编程 agent 端到端完成所有事情的原因。它们一直在变好。我们 Anthropic 有工程师已经不自己写任何代码了。当我看生产力时，回到你之前的问题，我们有人说，"这个 GPU kernel，这个芯片，我以前自己写的。现在我就让 Claude 来做。"生产力有巨大提升。当我看 Claude Code 时，对代码库的熟悉程度或者觉得模型没有在公司工作过一年，这些在我看到的抱怨列表中排名并不高。我觉得我要说的是，我们走的是一条不同的路。
**Dario Amodei:** Don't you think with coding that's because there is an external scaffold of memory which exists instantiated in the codebase? I don't know how many other jobs have that. Coding made fast progress precisely because it has this unique advantage that other economic activity doesn't.
**Dwarkesh Patel：** 你不觉得编程之所以进展快，恰恰是因为存在一个外部的记忆支架，它以代码库的形式存在吗？我不知道有多少其他工作有这个。编程之所以快速进步，恰恰是因为它有这个其他经济活动没有的独特优势。
**Dwarkesh Patel:** But when you say that, what you're implying is that by reading the codebase into the context, I have everything that the human needed to learn on the job. So that would be an example of—whether it's written or not, whether it's available or not—a case where everything you needed to know you got from the context window. What we think of as learning—"I started this job, it's going to take me six months to understand the code base"—the model just did it in the context. I honestly don't know how to think about this because there are people who qualitatively report what you're saying. I'm sure you saw last year, there was a major study where they had experienced developers try to close pull requests in repositories that they were familiar with. Those developers reported an uplift. They reported that they felt more productive with the use of these models. But in fact, if you look at their output and how much was actually merged back in, there was a 20% downlift. They were less productive as a result of using these models. So I'm trying to square the qualitative feeling that people feel with these models versus, 1) in a macro level, where is this renaissance of software? And then 2) when people do these independent evaluations, why are we not seeing the productivity benefits we would expect?
Within Anthropic, this is just really unambiguous. We're under an incredible amount of commercial pressure and make it even harder for ourselves because we have all this safety stuff we do that I think we do more than other companies. The pressure to survive economically while also keeping our values is just incredible. We're trying to keep this 10x revenue curve going. There is zero time for bullshit. There is zero time for feeling like we're productive when we're not. These tools make us a lot more productive. Why do you think we're concerned about competitors using the tools? Because we think we're ahead of the competitors. We wouldn't be going through all this trouble if this were secretly reducing our productivity. We see the end productivity every few months in the form of model launches. There's no kidding yourself about this. The models make you more productive.
Within Anthropic, this is just really unambiguous. We're under an incredible amount of commercial pressure and make it even harder for ourselves because we have all this safety stuff we do that I think we do more than other companies. The pressure to survive economically while also keeping our values is just incredible. We're trying to keep this 10x revenue curve going. There is zero time for bullshit. There is zero time for feeling like we're productive when we're not. These tools make us a lot more productive. Why do you think we're concerned about competitors using the tools? Because we think we're ahead of the competitors. We wouldn't be going through all this trouble if this were secretly reducing our productivity. We see the end productivity every few months in the form of model launches. There's no kidding yourself about this. The models make you more productive.
**Dario Amodei：** 但当你这么说的时候，你暗示的是通过把代码库读入上下文，我就拥有了人类在工作中需要学习的一切。所以这就是一个例子——不管它是不是写下来的，不管它是否可用——一个你需要知道的一切都来自上下文窗口的案例。我们认为的学习——"我开始这份工作，我需要六个月来理解代码库"——模型在上下文中就做到了。我老实说不知道怎么看待这个问题，因为确实有人在定性上报告了你说的那些。我相信你去年看到了，有一个大型研究，让有经验的开发者在他们熟悉的代码库中尝试关闭 pull request。那些开发者报告说有提升。他们报告说使用这些模型时感觉更有效率。但实际上，如果你看他们的产出和实际被合并回去的量，有 20% 的下降。他们使用这些模型后生产力反而降低了。所以我试图把人们使用这些模型时的主观感受和以下两点对照起来：1）在宏观层面，软件的复兴在哪里？2）当人们做这些独立评估时，为什么我们没有看到预期的生产力收益？
在 Anthropic 内部，这是毫不含糊的。我们承受着巨大的商业压力，而且因为我们做的安全工作比其他公司多，所以压力更大。在经济生存和坚持价值观之间保持平衡的压力是难以置信的。我们在努力保持这条 10 倍收入曲线。没有时间自欺欺人。没有时间假装有效率但实际上没有。这些工具确实让我们高效得多。你觉得我们为什么担心竞争对手使用这些工具？因为我们认为我们领先于竞争对手。如果这在暗中降低我们的生产力，我们不会费这么大的劲。我们每隔几个月就以模型发布的形式看到最终生产力。在这方面是骗不了自己的。模型确实让你更有效率。
在 Anthropic 内部，这是毫不含糊的。我们承受着巨大的商业压力，而且因为我们做的安全工作比其他公司多，所以压力更大。在经济生存和坚持价值观之间保持平衡的压力是难以置信的。我们在努力保持这条 10 倍收入曲线。没有时间自欺欺人。没有时间假装有效率但实际上没有。这些工具确实让我们高效得多。你觉得我们为什么担心竞争对手使用这些工具？因为我们认为我们领先于竞争对手。如果这在暗中降低我们的生产力，我们不会费这么大的劲。我们每隔几个月就以模型发布的形式看到最终生产力。在这方面是骗不了自己的。模型确实让你更有效率。
**Dario Amodei:** 1) People feeling like they're productive is qualitatively predicted by studies like this. But 2) if I just look at the end output, obviously you guys are making fast progress. But the idea was supposed to be that with recursive self-improvement, you make a better AI, the AI helps you build a better next AI, et cetera, et cetera. What I see instead—if I look at you, OpenAI, DeepMind—is that people are just shifting around the podium every few months. Maybe you think that stops because you've won or whatever. But why are we not seeing the person with the best coding model have this lasting advantage if in fact there are these enormous productivity gains from the last coding model.
**Dwarkesh Patel：** 1）人们觉得自己高效，这在定性上和这类研究是一致的。但 2）如果我只看最终产出，显然你们在快速进步。但本来的设想是递归自我改进——你做出一个更好的 AI，AI 帮你构建下一个更好的 AI，如此循环。但我看到的——如果我看你们、OpenAI、DeepMind——是人们每隔几个月就在领奖台上轮换位置。也许你认为这会停止因为你们赢了或什么的。但如果事实上最新的编程模型带来了这些巨大的生产力收益，为什么我们没看到拥有最好编程模型的人有这种持久的优势？
**Dwarkesh Patel:** I think my model of the situation is that there's an advantage that's gradually growing. I would say right now the coding models give maybe, I don't know, a 15-20% total factor speed up. That's my view. Six months ago, it was maybe 5%. So it didn't matter. 5% doesn't register. It's now just getting to the point where it's one of several factors that kind of matters. That's going to keep speeding up. I think six months ago, there were several companies that were at roughly the same point because this wasn't a notable factor, but I think it's starting to speed up more and more. I would also say there are multiple companies that write models that are used for code and we're not perfectly good at preventing some of these other companies from using our models internally. So I think everything we're seeing is consistent with this kind of snowball model. Again, my theme in all of this is all of this is soft takeoff, soft, smooth exponentials, although the exponentials are relatively steep. So we're seeing this snowball gather momentum where it's like 10%, 20%, 25%, 40%. As you go, Amdahl's law, you have to get all the things that are preventing you from closing the loop out of the way. But this is one of the biggest priorities within Anthropic.
**Dario Amodei：** 我认为我对局势的模型是，有一个逐渐增长的优势。我会说现在编程模型大概给出了，我不知道，15-20% 的总体速度提升。这是我的看法。六个月前，可能是 5%。5% 根本不会被注意到。它现在刚刚达到了一个程度，成为了几个有点重要的因素之一。这个速度还会继续加快。我认为六个月前，有几家公司大致处于同一水平，因为这不是一个显著的因素，但我认为它正在越来越快地加速。我还要说，有多家公司编写用于编程的模型，而且我们并不完全能阻止其中一些公司在内部使用我们的模型。所以我认为我们看到的一切都与这种滚雪球模型一致。再说一遍，我在所有这些中的主题是，所有这些都是软起飞，平滑的指数增长，虽然指数增长相对陡峭。所以我们看到这个雪球在积聚势头，比如 10%、20%、25%、40%。随着你的推进，Amdahl 定律，你必须把阻碍你闭环的所有东西清除掉。但这是 Anthropic 内部最大的优先事项之一。
**Dario Amodei:** Stepping back, before in the stack we were talking about when do we get this on-the-job learning? It seems like the point you were making on the coding thing is that we actually don't need on-the-job learning. You can have tremendous productivity improvements, you can have potentially trillions of dollars of revenue for AI companies, without this basic human ability to learn on the job. Maybe that's not your claim, you should clarify. But in most domains of economic activity, people say, "I hired somebody, they weren't that useful for the first few months, and then over time they built up the context, understanding." It's actually hard to define what we're talking about here. But they got something and then now they're a powerhorse and they're so valuable to us. If AI doesn't develop this ability to learn on the fly, I'm a bit skeptical that we're going to see huge changes to the world without that ability.
**Dwarkesh Patel：** 退一步来看，在整个对话中，我们之前一直在讨论什么时候能获得在职学习能力。你在编程这个话题上要表达的观点似乎是，我们实际上不需要在职学习。你可以获得巨大的生产力提升，AI 公司可能有数万亿美元的收入，而不需要这种基本的人类在职学习能力。也许这不是你的声明，你应该澄清一下。但在大多数经济活动领域，人们会说，"我雇了一个人，他们头几个月不太有用，然后随着时间推移，他们建立了上下文和理解。"这实际上很难界定我们在说什么。但他们获得了某些东西，然后现在他们成了公司的顶梁柱，对我们非常有价值。如果 AI 不发展出这种即时学习的能力，我对我们能否看到世界的巨大变化持怀疑态度。
**Dwarkesh Patel:** I think two things here. There's the state of the technology right now. Again, we have these two stages. We have the pre-training and RL stage where you throw a bunch of data and tasks into the models and then they generalize. So it's like learning, but it's like learning from more data and not learning over one human or one model's lifetime. So again, this is situated between evolution and human learning. But once you learn all those skills, you have them. Just like with pre-training, just how the models know more, if I look at a pre-trained model, it knows more about the history of samurai in Japan than I do. It knows more about baseball than I do. It knows more about low-pass filters and electronics, all of these things. Its knowledge is way broader than mine. So I think even just that may get us to the point where the models are better at everything. We also have, again, just with scaling the kind of existing setup, the in-context learning. I would describe it as kind of like human on-the-job learning, but a little weaker and a little short term. You look at in-context learning and if you give the model a bunch of examples it does get it. There's real learning that happens in context. A million tokens is a lot. That can be days of human learning. If you think about the model reading a million words, how long would it take me to read a million? Days or weeks at least. So you have these two things. I think these two things within the existing paradigm may just be enough to get you the "country of geniuses in a data center". I don't know for sure, but I think they're going to get you a large fraction of it. There may be gaps, but I certainly think that just as things are, this is enough to generate trillions of dollars of revenue. That's one. Two, is this idea of continual learning, this idea of a single model learning on the job. I think we're working on that too. There's a good chance that in the next year or two, we also solve that. Again, I think you get most of the way there without it. The trillions of dollars a year market, maybe all of the national security implications and the safety implications that I wrote about in "Adolescence of Technology" can happen without it. But we, and I imagine others, are working on it. There's a good chance that we will get there within the next year or two. There are a bunch of ideas. I won't go into all of them in detail, but one is just to make the context longer. There's nothing preventing longer contexts from working. You just have to train at longer contexts and then learn to serve them at inference. Both of those are engineering problems that we are working on and I would assume others are working on them as well.
**Dario Amodei：** 我觉得这里有两件事。首先是技术的当前状态。我们有两个阶段。我们有预训练和 RL 阶段，你把一堆数据和任务扔给模型，然后它们泛化。所以这像学习，但是从更多数据中学习，而不是在一个人或一个模型的生命周期内学习。所以再说一遍，这位于进化和人类学习之间。但一旦你学会了所有这些技能，你就拥有了它们。就像预训练一样，模型知道更多东西。如果我看一个预训练模型，它比我更了解日本武士的历史。它比我更了解棒球。它比我更了解低通滤波器和电子学，所有这些东西。它的知识广度远超过我。所以我认为即使只是这些，也可能让模型在所有事情上都比人类强。我们还有，同样是通过扩展现有架构，上下文学习。我会把它描述为有点像人类的在职学习，但稍微弱一些，稍微短期一些。你看上下文学习，如果你给模型一堆例子，它确实能学会。上下文中确实有真正的学习在发生。一百万个 token 是很多的。那可以是人类好几天的学习量。如果你想想模型读一百万词，我读一百万词需要多长时间？至少几天或几周。所以你有这两样东西。我认为在现有范式内，这两样东西可能就足以让你得到"数据中心中的天才之国"。我不确定，但我认为它们会让你获得很大一部分。可能有差距，但我确信仅凭现状，这足以产生数万亿美元的收入。这是第一点。第二点，是持续学习的概念，单个模型在职学习的概念。我认为我们也在研究这个。在未来一两年内，我们也有很大机会解决它。再说一遍，我认为没有它你也能走完大部分路程。每年数万亿美元的市场，也许我在"Adolescence of Technology"中写到的所有国家安全影响和安全影响，都可以在没有它的情况下发生。但我们——我想象其他人也是——在研究这个。在未来一两年内，我们有很大机会达到那里。有很多想法。我不会一一详述，但其中一个就是把上下文长度加长。没有什么阻止更长的上下文长度起作用。你只需要在更长的上下文上训练，然后在推理时学会服务它们。这两者都是工程问题，我们正在研究，我猜其他人也在研究。
**Dario Amodei:** This context length increase, it seemed like there was a period from 2020 to 2023 where from GPT-3 to GPT-4 Turbo, there was an increase from 2000 context lengths to 128K. I feel like for the two-ish years since then, we've been in the same-ish ballpark. When context lengths get much longer than that, people report qualitative degradation in the ability of the model to consider that full context. So I'm curious what you're internally seeing that makes you think, "10 million contexts, 100 million contexts to get six months of human learning and building context".
**Dwarkesh Patel：** 这个上下文长度的增长，似乎从 2020 到 2023 年有一段时期，从 GPT-3 到 GPT-4 Turbo，从 2000 的上下文长度增长到了 128K。我感觉从那以后的大约两年里，我们一直在差不多的范围内。当上下文长度远超这个范围时，人们报告说模型考虑完整上下文的能力出现了质的下降。所以我好奇你在内部看到了什么，让你觉得"一千万上下文、一亿上下文能获得六个月的人类学习和上下文积累"。
**Dwarkesh Patel:** This isn't a research problem. This is an engineering and inference problem. If you want to serve long context, you have to store your entire KV cache. It's difficult to store all the memory in the GPUs, to juggle the memory around. I don't even know the details. At this point, this is at a level of detail that I'm no longer able to follow, although I knew it in the GPT-3 era. "These are the weights, these are the activations you have to store…" But these days the whole thing is flipped because we have MoE models and all of that. Regarding this degradation you're talking about, without getting too specific, there's two things. There's the context length you train at and there's a context length that you serve at. If you train at a small context length and then try to serve at a long context length, maybe you get these degradations. It's better than nothing, you might still offer it, but you get these degradations. Maybe it's harder to train at a long context length.
**Dario Amodei：** 这不是一个研究问题。这是一个工程和推理问题。如果你想服务长上下文，你必须存储整个 KV cache。在 GPU 中存储所有内存、调度内存是困难的。我甚至不知道具体细节。到了这个层面的细节，我已经跟不上了，虽然在 GPT-3 时代我是知道的。"这些是权重，这些是你需要存储的激活值……"但现在整个事情已经翻转了，因为我们有了 MoE 模型和所有这些。关于你说的退化，在不说得太具体的情况下，有两件事。有训练时的上下文长度，有服务时的上下文长度。如果你在短上下文长度上训练然后试图在长上下文长度上服务，也许你会得到这些退化。聊胜于无，你可能还是会提供它，但你确实会得到退化。也许在长上下文长度上训练更难。
**Dario Amodei:** I want to, at the same time, ask about maybe some rabbit holes. Wouldn't you expect that if you had to train on longer context length, that would mean that you're able to get less samples in for the same amount of compute? Maybe it's not worth diving deep on that. I want to get an answer to the bigger picture question. I don't feel a preference for a human editor that's been working for me for six months versus an AI that's been working with me for six months, what year do you predict that that will be the case?
**Dwarkesh Patel：** 我想同时问一下，也许是一些兔子洞。你不觉得如果你必须在更长的上下文长度上训练，那意味着在同样的计算量下你能获得更少的样本吗？也许不值得深入探讨这个。我想得到一个更大图景的答案。我不再偏好一个为我工作了六个月的人类编辑而非一个和我一起工作了六个月的 AI，你预测这会在哪一年成为现实？
**Dwarkesh Patel:** My guess for that is there's a lot of problems where basically we can do this when we have the "country of geniuses in a data center". My picture for that, if you made me guess, is one to two years, maybe one to three years. It's really hard to tell. I have a strong view—99%, 95%—that all this will happen in 10 years. I think that's just a super safe bet. I have a hunch—this is more like a 50/50 thing—that it's going to be more like one to two, maybe more like one to three. So one to three years.
**Dario Amodei：** 我的猜测是，有很多问题基本上当我们拥有"数据中心中的天才之国"时就能解决。我对此的预测，如果你让我猜的话，是一到两年，也许一到三年。真的很难说。我有一个强烈的观点——99%、95%——所有这些都会在十年内发生。我认为这是一个超级安全的赌注。我有一个直觉——这更像是 50/50 的事情——它会更像一到两年，也许更像一到三年。
**Dario Amodei:** Country of geniuses, and the slightly less economically valuable task of editing videos. It seems pretty economically valuable, let me tell you. It's just there are a lot of use cases like that. There are a lot of similar ones. So you're predicting that within one to three years. And then, generally, Anthropic has predicted that by late '26 or early '27 we will have AI systems that "have the ability to navigate interfaces available to humans doing digital work today, intellectual capabilities matching or exceeding that of Nobel Prize winners, and the ability to interface with the physical world". You gave an interview two months ago with DealBook where you were emphasizing your company's more responsible compute scaling as compared to your competitors. I'm trying to square these two views. If you really believe that we're going to have a country of geniuses, you want as big a data center as you can get. There's no reason to slow down. The TAM of a Nobel Prize winner, that can actually do everything a Nobel Prize winner can do, is trillions of dollars. So I'm trying to square this conservatism, which seems rational if you have more moderate timelines, with your stated views about progress.
**Dwarkesh Patel：** 所以一到三年。天才之国，以及稍微经济价值低一些的视频编辑任务。
**Dwarkesh Patel:** It actually all fits together. We go back to this fast, but not infinitely fast, diffusion. Let's say that we're making progress at this rate. The technology is making progress this fast. I have very high conviction that we're going to get there within a few years. I have a hunch that we're going
**Dario Amodei：** 视频编辑看起来相当有经济价值，让我告诉你。只是有很多类似的用例。有很多类似的。
**Dario Amodei:** So one to three years. Country of geniuses, and the slightly less economically valuable task of editing videos.
**Dwarkesh Patel：** 所以你预测在一到三年内。然后，总的来说，Anthropic 预测到 2026 年底或 2027 年初，我们将拥有"具备在人类进行数字工作时所使用界面上导航的能力、智力能力匹配或超过诺贝尔奖获得者、以及与物理世界交互的能力"的 AI 系统。你两个月前在 DealBook 的采访中强调你们公司相比竞争对手在算力扩张上更负责任。我试图把这两个观点协调起来。如果你真的相信我们会有一个天才之国，你会想要尽可能大的数据中心。没有理由放慢。一个真正能做诺贝尔奖获得者所能做的一切的人，TAM 是数万亿美元。所以我试图把这种保守主义——在更温和的时间线下看起来是理性的——与你对进展的公开表态协调起来。
**Dwarkesh Patel:** It seems pretty economically valuable, let me tell you. It's just there are a lot of use cases like that. There are a lot of similar ones.
**Dario Amodei：** 这实际上全部是一致的。我们回到这个快速但不是无限快的扩散。假设我们正在以这个速率取得进展。技术正在这么快地进步。我非常有信心我们会在几年内到达那里。我有一个直觉，我们会在一两年内到达。所以技术方面有一点不确定性，但相当确信不会差太多。我不太确定的是经济扩散方面。我真的相信我们可以在一到两年内拥有作为数据中心中天才之国的模型。一个问题是：在那之后多少年万亿美元的收入才会开始涌入？我不认为这一定是立即的。可能是一年，可能是两年，我甚至可以延伸到五年，虽然我对此持怀疑态度。所以我们有这种不确定性。即使技术以我怀疑的速度发展，我们也不确切知道它会多快推动收入。我们知道它会来，但按照你购买数据中心的方式，如果你偏差了几年，那可能是灾难性的。就像我在"Machines of Loving Grace"中写的那样。我说我认为我们可能会得到这种强大的 AI，这个"数据中心中的天才之国"。你给出的描述就来自"Machines of Loving Grace"。我说我们会在 2026 年得到它，也许 2027 年。再说一遍，这是我的直觉。如果我偏差了一两年我不会惊讶，但这是我的直觉。假设这发生了。那就是起跑枪。治愈所有疾病需要多长时间？这是驱动大量经济价值的方式之一。你治愈每一种疾病。制药公司或 AI 公司各得多少是一个问题，但有巨大的消费者剩余，因为——假设我们能让每个人都获得，这是我非常关心的——我们治愈了所有这些疾病。需要多长时间？你必须做生物学发现，你必须生产新药，你必须经过监管流程。我们在疫苗和 COVID 上看到了这一点。我们把疫苗送到了每个人手中，但花了一年半。我的问题是：从那个 AI 首次存在于实验室中，到疾病真正为所有人治愈，需要多长时间？我们已经有脊髓灰质炎疫苗 50 年了。我们仍在努力在非洲最偏远的角落根除它。Gates Foundation 在尽其所能。其他人也在尽其所能。但那很困难。再说一遍，我不预期大多数经济扩散会像那样困难。那是最困难的情况。但这里有一个真正的困境。我的结论是，它会比我们在世界上见过的任何东西都快，但仍然有其极限。所以当我们去购买数据中心时，我看的曲线是：我们每年都有 10 倍的增长。今年初，我们看到的是 100 亿美元的年化收入。我们必须决定买多少算力。实际建设数据中心、预留数据中心需要一两年。基本上我在说，"2027 年我能获得多少算力？"我可以假设收入会继续每年 10 倍增长，所以 2026 年底是 1000 亿美元，2027 年底是 1 万亿美元。实际上如果算五年的话，那就是 5 万亿美元的算力，因为每年 1 万亿持续五年。我可以购买从 2027 年底开始的 1 万亿美元的算力。如果我的收入不是 1 万亿美元，如果哪怕只是 8000 亿美元，世界上没有任何力量、没有任何对冲能阻止我破产，如果我买了那么多算力的话。即使我脑中的一部分想知道它是否会继续 10 倍增长，我也不能在 2027 年买每年 1 万亿美元的算力。如果那个增长率只偏了一年，或者增长率是每年 5 倍而不是 10 倍，那你就破产了。所以你最终进入一个世界，你支撑数千亿，而不是数万亿。你接受一些风险，即需求太大你无法支撑收入，同时接受一些风险，即你搞错了，进展仍然很慢。当我谈到负责任行事时，我实际上指的不是绝对金额。我认为确实我们花得比一些其他玩家少一些。实际上是其他方面——比如我们是否经过深思熟虑，还是在瞎搞说"我们在这里投 1000 亿或在那里投 1000 亿"？我的印象是一些其他公司没有做过电子表格，他们并不真正理解自己承担的风险。他们只是因为听起来很酷就在做。我们仔细想过了。我们是一家企业级公司。因此，我们可以更多地依赖收入。它没有消费者业务那么不稳定。我们有更好的利润率，这是在买太多和买太少之间的缓冲。我认为我们购买的数量让我们能够捕获相当强劲的上行情景。它不会捕获完整的每年 10 倍。事情必须相当糟糕我们才会陷入财务困境。所以我们仔细想过了，做了那个平衡。这就是我说我们在负责任行事时的意思。
**Dario Amodei:** So you're predicting that within one to three years. And then, generally, Anthropic has predicted that by late '26 or early '27 we will have AI systems that "have the ability to navigate interfaces available to humans doing digital work today, intellectual capabilities matching or exceeding that of Nobel Prize winners, and the ability to interface with the physical world". You gave an interview two months ago with DealBook where you were emphasizing your company's more responsible compute scaling as compared to your competitors. I'm trying to square these two views. If you really believe that we're going to have a country of geniuses, you want as big a data center as you can get. There's no reason to slow down. The TAM of a Nobel Prize winner, that can actually do everything a Nobel Prize winner can do, is trillions of dollars. So I'm trying to square this conservatism — which seems rational if you have more moderate timelines — with your stated views about progress.
**Dwarkesh Patel：** 所以似乎我们实际上可能只是对"数据中心中的天才之国"有不同的定义。因为当我想到真正的人类天才，一个真正的人类天才之国在数据中心里，我会很乐意花 5 万亿美元的算力来运行一个真正的人类天才之国在数据中心里。假设 JPMorgan 或 Moderna 或什么公司不想用它们。我有一个天才之国。它们会自己创业。如果它们不能自己创业，被临床试验卡住了……
**Dwarkesh Patel:** It actually all fits together. We go back to this fast, but not infinitely fast, diffusion. Let's say that we're making progress at this rate. The technology is making progress this fast. I have very high conviction that we're going to get there within a few years. I have a hunch that we're going to get there within a year or two. So there's a little uncertainty on the technical side, but pretty strong confidence that it won't be off by much. What I'm less certain about is, again, the economic diffusion side. I really do believe that we could have models that are a country of geniuses in the data center in one to two years. One question is: How many years after that do the trillions in revenue start rolling in? I don't think it's guaranteed that it's going to be immediate. It could be one year, it could be two years, I could even stretch it to five years although I'm skeptical of that. So we have this uncertainty. Even if the technology goes as fast as I suspect that it will, we don't know exactly how fast it's going to drive revenue. We know it's coming, but with the way you buy these data centers, if you're off by a couple years, that can be ruinous. It is just like how I wrote in "Machines of Loving Grace". I said I think we might get this powerful AI, this "country of genius in the data center". That description you gave comes from "Machines of Loving Grace". I said we'll get that in 2026, maybe 2027. Again, that is my hunch. I wouldn't be surprised if I'm off by a year or two, but that is my hunch. Let's say that happens. That's the starting gun. How long does it take to cure all the diseases? That's one of the ways that drives a huge amount of economic value. You cure every disease. There's a question of how much of that goes to the pharmaceutical company or the AI company, but there's an enormous consumer surplus because — assuming we can get access for everyone, which I care about greatly — we cure all of these diseases. How long does it take? You have to do the biological discovery, you have to manufacture the new drug, you have to go through the regulatory process. We saw this with vaccines and COVID. We got the vaccine out to everyone, but it took a year and a half. My question is: How long does it take to get the cure for everything — which AI is the genius that can in theory invent — out to everyone? How long from when that AI first exists in the lab to when diseases have actually been cured for everyone? We've had a polio vaccine for 50 years. We're still trying to eradicate it in the most remote corners of Africa. The Gates Foundation is trying as hard as they can. Others are trying as hard as they can. But that's difficult. Again, I don't expect most of the economic diffusion to be as difficult as that.
**Dario Amodei：** 值得指出的是，对于临床试验，大多数临床试验失败是因为药物无效。缺乏疗效。我在"Machines of Loving Grace"中恰好提出了这一点。我说临床试验会比我们习惯的快得多，但不是无限快。
**Dario Amodei:** That's the most difficult case. But there's a real dilemma here. Where I've settled on it is that it will be faster than anything we've seen in the world, but it still has its limits. So when we go to buying data centers, again, the curve I'm looking at is: we've had a 10x a year increase every year. At the beginning of this year, we're looking at $10 billion in annualized revenue. We have to decide how much compute to buy. It takes a year or two to actually build out the data centers, to reserve the data center. Basically I'm saying, "In 2027, how much compute do I get?" I could assume that the revenue will continue growing 10x a year, so it'll be $100 billion at the end of 2026 and $1 trillion at the end of 2027. Actually it would be $5 trillion dollars of compute because it would be $1 trillion a year for five years. I could buy $1 trillion of compute that starts at the end of 2027. If my revenue is not $1 trillion dollars, if it's even $800 billion, there's no force on earth, there's no hedge on earth that could stop me from going bankrupt if I buy that much compute. Even though a part of my brain wonders if it's going to keep growing 10x, I can't buy $1 trillion a year of compute in 2027. If I'm just off by a year in that rate of growth, or if the growth rate is 5x a year instead of 10x a year, then you go bankrupt. So you end up in a world where you're supporting hundreds of billions, not trillions. You accept some risk that there's so much demand that you can't support the revenue, and you accept some risk that you got it wrong and it's still slow. When I talked about behaving responsibly, what I meant actually was not the absolute amount. I think it is true we're spending somewhat less than some of the other players. It's actually the other things, like have we been thoughtful about it or are we YOLOing and saying, "We're going to do $100 billion here or $100 billion there"? I get the impression that some of the other companies have not written down the spreadsheet, that they don't really understand the risks they're taking. They're just doing stuff because it sounds cool. We've thought carefully about it. We're an enterprise business. Therefore, we can rely more on revenue. It's less fickle than consumer. We have better margins, which is the buffer between buying too much and buying too little. I think we bought an amount that allows us to capture pretty strong upside worlds. It won't capture the full 10x a year. Things would have to go pretty badly for us to be in financial trouble. So we've thought carefully and we've made that balance. That's what I mean when I say that we're being responsible.
**Dwarkesh Patel：** 好吧，假设临床试验需要一年才能得到结果，然后你就能从中获得收入，制造更多药物。好的，你有一个天才之国，你是一家 AI 实验室。你可以用更多 AI 研究员。你也认为聪明人在 AI 技术上工作有这些自我强化的收益。你可以让数据中心致力于 AI 进展。购买每年 1 万亿美元的算力比每年 3000 亿美元的算力有实质性的更多收益吗？
**Dwarkesh Patel:** So it seems like it's possible that we actually just have different definitions of the "country of a genius in a data center". Because when I think of actual human geniuses, an actual country of human geniuses in a data center, I would happily buy $5 trillion worth of compute to run an actual country of human geniuses in a data center. Let's say JPMorgan or Moderna or whatever doesn't want to use them. I've got a country of geniuses. They'll start their own company. If they can't start their own company and they're bottlenecked by clinical trials…
**Dario Amodei：** 如果你的竞争对手在购买一万亿，那是的。
**Dario Amodei:** It is worth stating that with clinical trials, most clinical trials fail because the drug doesn't work. There's not efficacy. I make exactly that point in "Machines of Loving Grace", I say the clinical trials are going to go much faster than we're used to, but not infinitely fast.
**Dwarkesh Patel：** 好吧，不，有一些收益，但话说回来，还有他们之前就破产的可能。
**Dwarkesh Patel:** Okay, and then suppose it takes a year for the clinical trials to work out so that you're getting revenue from that and can make more drugs. Okay, well, you've got a country of geniuses and you're an AI lab. You could use many more AI researchers. You also think there are these self-reinforcing gains from smart people working on AI tech. You can have the data center working on AI progress. Are there substantially more gains from buying $1 trillion a year of compute versus $300 billion a year of compute?
**Dario Amodei：** 再说一遍，如果你只偏了一年，你就毁了自己。这就是那个平衡。我们买了很多。我们买了非常非常多。我们买的量和这个行业最大的玩家买的量差不多。但如果你问我，"为什么你们没有签下从 2027 年中开始的 10 万亿美元的算力？"……首先，生产不出来。世界上没有那么多。但其次，如果天才之国是在 2028 年中而不是 2027 年中到来呢？你就破产了。
**Dario Amodei:** If your competitor is buying a trillion, yes there is.
**Dwarkesh Patel：** 所以如果你的预测是一到三年，看起来你应该想在最晚 2029 年之前拥有 10 万亿美元的算力？即使在你说的最长版本的时间线中，你正在加速建设的算力似乎与之不符。是什么让你这么想？人类工资，比方说，大约每年 50 万亿美元——
**Dwarkesh Patel:** Well, no, there's some gain, but then again, there's this chance that they go bankrupt before.
**Dario Amodei：** 所以我不会谈 Anthropic 的具体情况，但如果你谈整个行业，这个行业今年正在建设的算力大概是 10-15 吉瓦。它每年大约增长 3 倍。所以明年是 30-40 吉瓦。2028 年可能是 100 吉瓦。2029 年可能大约是 300 吉瓦。我在心里算，但每吉瓦大约花费 100 亿美元，每年大约 100-150 亿美元。把这些加在一起，你得到的大约就是你描述的。你得到的正是那个数字。到 2028 或 2029 年，你得到的是每年数万亿。你得到的正是你预测的。
**Dario Amodei:** Again, if you're off by only a year, you destroy yourselves. That's the balance. We're buying a lot. We're buying a hell of a lot. We're buying an amount that's comparable to what the biggest players in the game are buying. But if you're asking me, "Why haven't we signed $10 trillion of compute starting in mid-2027?"... First of all, it can't be produced. There isn't that much in the world. But second, what if the country of geniuses comes, but it comes in mid-2028 instead of mid-2027? You go bankrupt.
**Dwarkesh Patel：** 那是整个行业的。
**Dwarkesh Patel:** So if your projection is one to three years, it seems like you should want $10 trillion of compute by 2029 at the latest? Even in the longest version of the timelines you state, the compute you are ramping up to build doesn't seem in accordance. What makes you think that? Human wages, let's say, are on the order of $50 trillion a year —
**Dario Amodei：** 那是整个行业的，对。假设 Anthropic 的算力继续每年 3 倍增长，然后到 2027-28 年，你有 10 吉瓦。乘以你说的 100 亿美元。那就是大约每年 1000 亿美元。但你说的 2028 年 TAM 是 2000 亿美元。
**Dario Amodei:** So I won't talk about Anthropic in particular, but if you talk about the industry, the amount of compute the industry is building this year is probably, call it, 10-15 gigawatts. It goes up by roughly 3x a year. So next year's 30-40 gigawatts. 2028 might be 100 gigawatts. 2029 might be like 300 gigawatts. I'm doing the math in my head, but each gigawatt costs maybe $10 billion, on the order of $10-15 billion a year. You put that all together and you're getting about what you described. You're getting exactly that. You're getting multiple trillions a year by 2028 or 2029. You're getting exactly what you predict.
**Dwarkesh Patel：** 再说一遍，我不想给 Anthropic 的确切数字，但这些数字太小了。
**Dwarkesh Patel:** That's for the industry.
**Dario Amodei：** 好的，有意思。
**Dario Amodei:** That's for the industry, that's right. Suppose Anthropic's compute keeps 3x-ing a year, and then by 2027-28, you have 10 gigawatts. Multiply that by, as you say, $10 billion. So then it's like $100 billion a year. But then you're saying the TAM by 2028 is $200 billion.
**Dwarkesh Patel：** 你告诉投资者你计划从 2028 年开始盈利。这正是我们可能获得数据中心天才之国的年份。这将解锁医学、健康和新技术的所有进展。这难道不正是你想要再投资业务、建造更大"国家"让它们做更多发现的时候吗？
**Dwarkesh Patel:** Again, I don't want to give exact numbers for Anthropic, but these numbers are too small.
**Dario Amodei：** 盈利在这个领域是一个有点奇怪的事情。我不认为在这个领域，盈利性实际上是在衡量减少支出还是投资业务。让我用一个模型来说。我实际上认为盈利发生在你低估了将要获得的需求量的时候，而亏损发生在你高估了将要获得的需求量的时候，因为你是提前购买数据中心的。这样想。再说一遍，这些是简化的事实。这些数字不是精确的。我只是在试图做一个简化模型。假设你一半的算力用于训练，一半用于推理。推理有一个超过 50% 的毛利率。这意味着如果你处于稳态，你建了一个数据中心，如果你确切地知道要获得的需求，你会得到一定量的收入。假设你每年在算力上花 1000 亿美元。在其中 500 亿上你支撑了 1500 亿的收入。另外 500 亿用于训练。基本上你是盈利的，你赚 500 亿利润。这就是今天这个行业的经济模型，或者不是今天，而是我们在一两年后预测的。唯一让这不成立的是如果你获得的需求少于 500 亿。那你就有超过 50% 的数据中心用于研究，你不盈利。所以你训练更强的模型，但你不盈利。如果你获得的需求超出预期，那研究就被挤压了，但你能支撑更多推理，你更盈利。也许我没解释好，但我想说的是，你首先决定算力的量。然后你有一些推理与训练的目标分配，但那是由需求决定的。不是由你决定的。
**Dario Amodei:** Okay, interesting.
**Dwarkesh Patel：** 我听到的是，你预测盈利的原因是你在算力上系统性地投资不足？
**Dwarkesh Patel:** You've told investors that you plan to be profitable starting in 2028. This is the year when we're potentially getting the country of geniuses as a data center. This is now going to unlock all this progress in medicine and health and new technologies. Wouldn't this be exactly the time where you'd want to reinvest in the business and build bigger "countries" so they can make more discoveries?
**Dario Amodei：** 不，不，不。我是说这很难预测。关于 2028 年和什么时候会发生这些事情，那是我们尽最大努力向投资者说明的。所有这些都非常不确定，因为不确定性的锥形范围。如果收入增长够快，我们可能 2026 年就盈利了。如果我们对来年高估或低估，那可能会剧烈波动。我想说的是，你脑中有一个投资、投资、投资，达到规模然后盈利的商业模型。有一个单一的转折点。我不认为这个行业的经济模式是那样运作的。
**Dario Amodei:** Profitability is this kind of weird thing in this field. I don't think in this field profitability is actually a measure of spending down versus investing in the business. Let's just take a model of this. I actually think profitability happens when you underestimated the amount of demand you were going to get and loss happens when you overestimated the amount of demand you were going to get, because you're buying the data centers ahead of time. Think about it this way. Again, these are stylized facts. These numbers are not exact. I'm just trying to make a toy model here. Let's say half of your compute is for training and half of your compute is for inference. The inference has some gross margin that's more than 50%. So what that means is that if you were in steady-state, you build a data center and if you knew exactly the demand you were getting, you would get a certain amount of revenue. Let's say you pay $100 billion a year for compute. On $50 billion a year you support $150 billion of revenue. The other $50 billion is used for training. Basically you're profitable and you make $50 billion of profit. Those are the economics of the industry today, or not today but where we're projecting forward in a year or two. The only thing that makes that not the case is if you get less demand than $50 billion. Then you have more than 50% of your data center for research and you're not profitable. So you train stronger models, but you're not profitable. If you get more demand than you thought, then research gets squeezed, but you're kind of able to support more inference and you're more profitable. Maybe I'm not explaining it well, but the thing I'm trying to say is that you decide the amount of compute first. Then you have some target desire of inference versus training, but that gets determined by demand. It doesn't get determined by you.
**Dwarkesh Patel：** 我明白了。如果我理解正确的话，你是在说由于我们应该获得的算力和实际获得的算力之间的差异，我们被迫盈利了。但这不意味着我们会继续盈利。我们会再投资这笔钱，因为现在 AI 取得了这么大进步，我们想要更大的天才之国。所以又回到了收入很高但亏损也很高的状态。
**Dwarkesh Patel:** What I'm hearing is the reason you're predicting profit is that you are systematically underinvesting in compute?
**Dario Amodei：** 如果我们每年都准确预测需求，我们每年都会盈利。因为把 50% 的算力花在研究上，大致上，加上超过 50% 的毛利率和正确的需求预测，就会导致盈利。这就是我认为存在但被这些提前建设和预测误差掩盖的盈利商业模式。
**Dario Amodei:** No, no, no. I'm saying it's hard to predict. These things about 2028 and when it will happen, that's our attempt to do the best we can with investors. All of this stuff is really uncertain because of the cone of uncertainty. We could be profitable in 2026 if the revenue grows fast enough. If we overestimate or underestimate the next year, that could swing wildly. What I'm trying to get at is that you have a model in your head of a business that invests, invests, invests, gets scale and then becomes profitable. There's a single point at which things turn around. I don't think the economics of this industry work that way.
**Dwarkesh Patel：** 我猜你是把 50% 当作一种给定常量，而实际上，如果 AI 进展很快，你通过扩大规模可以增加进展，你就应该超过 50% 并且不盈利。
**Dwarkesh Patel:** I see. So if I'm understanding correctly, you're saying that because of the discrepancy between the amount of compute we should have gotten and the amount of compute we got, we were sort of forced to make profit. But that doesn't mean we're going to continue making profit. We're going to reinvest the money because now AI has made so much progress and we want a bigger country of geniuses. So back into revenue is high, but losses are also high.
**Dario Amodei：** 但我要说这个。你可能想扩大规模更多。记住对数收益递减。如果 70% 只会让你通过 1.4 倍的因子得到一个稍微小一点的模型……那额外的 200 亿美元，每一美元对你来说价值都小得多，因为是对数线性的。所以你可能发现把那 200 亿美元投资在服务推理或雇佣更擅长工作的工程师上更好。所以我说 50% 的原因……那不是我们的确切目标。不会正好是 50%。它可能会随时间变化。我要说的是，对数线性收益导致的是你在业务中花费大约一个分数。不是 5%，也不是 95%。然后你得到递减收益。
**Dario Amodei:** If every year we predict exactly what the demand is going to be, we'll be profitable every year. Because spending 50% of your compute on research, roughly, plus a gross margin that's higher than 50% and correct demand prediction leads to profit. That's the profitable business model that I think is kind of there, but obscured by these building ahead and prediction errors.
**Dwarkesh Patel：** 我觉得很奇怪，是我在说服 Dario 相信 AI 进展还是什么。好吧，你不投资研究因为它有递减收益，但你投资你提到的其他事情。我认为在宏观层面上的利润——
**Dwarkesh Patel:** I guess you're treating the 50% as a sort of given constant, whereas in fact, if AI progress is fast and
**Dario Amodei：** 再说一遍，我说的是递减收益，但那是在你已经每年花 500 亿之后。这是一个我相信你也会提出的观点，但天才身上的递减收益可能相当高。更一般地说，在市场经济中利润是什么？利润基本上是在说市场中的其他公司能用这笔钱做比我更多的事情。把 Anthropic 放在一边。我不想透露 Anthropic 的信息。这就是为什么我给这些简化的数字。但让我们推导一下行业的均衡。为什么不是每个人都把 100% 的算力花在训练上不服务任何客户？因为如果他们得不到任何收入，他们就不能融资，他们就不能做算力交易，他们明年就不能买更多算力。所以会有一个均衡，每家公司花不到 100% 在训练上，也肯定不到 100% 在推理上。为什么你不只服务当前模型而永远不训练另一个模型，这一点应该很清楚，因为那样你就没有需求了，因为你会落后。所以有某种均衡。它不会是 10%，也不会是 90%。就把它作为一个简化事实，假设是 50%。这就是我要说的。我认为我们会处于一个位置，在训练上花多少的均衡小于你在算力上能获得的毛利率。所以底层经济是盈利的。问题是你有这个地狱般的需求预测问题，当你购买下一年的算力时，你可能猜低了就非常盈利但没有算力做研究。或者你可能猜高了就不盈利但有全世界的算力做研究。这说得通吗？作为一个行业的动态模型？
**Dario Amodei:** you can increase the progress by scaling up more, you should just have more than 50% and not make profit.
**Dwarkesh Patel：** 也许退一步来看，我不是在说我认为"天才之国"两年后就来了所以你应该买这些算力。对我来说，你得出的最终结论很有道理。但那是因为"天才之国"看起来很难，还有很长的路要走。所以退一步来看，我想说的是，你的世界观似乎与那些说"我们距离产生数万亿美元价值的世界还有大约 10 年"的人是兼容的。
**Dwarkesh Patel:** I don't believe the economy is gonna grow 300% a year. I said this in "Machines of Loving Grace", I think we may get 10-20% per year growth in the economy, but we're not gonna get 300% growth in the economy.
**Dario Amodei：** 那根本不是我的观点。所以我再做一个预测。我很难想象在 2030 年之前不会有数万亿美元的收入。我能构建一个合理的世界。大概需要三年。那会是我认为合理的最慢端。比如 2028 年，我们得到了真正的"数据中心中的天才之国"。到 2028 年收入进入低数千亿，然后天才之国加速它到数万亿。我们基本上处于扩散的慢端。到达数万亿需要两年。那就是需要到 2030 年的世界。我怀疑即使把技术指数和扩散指数组合在一起，我们也会在 2030 年之前到达那里。
**Dario Amodei:** So I think in the end, if compute becomes the majority of what the economy produces, it's gonna be capped by that.
**Dwarkesh Patel：** 所以你铺开了一个模型，Anthropic 盈利是因为我们从根本上处于一个算力受限的世界。所以最终我们继续增长算力——
**Dwarkesh Patel:** So let's assume a model where compute stays capped. The world where frontier labs are making money is one where they continue to make fast progress. Because fundamentally your margin is limited by how good the alternative is. So you are able to make money because you have a frontier model. If you didn't have a frontier model you wouldn't be making money. So this model requires there never to be a steady state. Forever and ever you keep making more algorithmic progress.
**Dario Amodei：** 我认为利润的来源是……再说一遍，让我们把整个行业抽象化。假设我们在一本经济学教科书里。我们有少数几家公司。每家可以投资有限的量。每家可以把一些比例投资在研发上。它们有一些边际服务成本。推理效率高所以毛利率很高。有一些竞争，但模型也是差异化的。公司会竞争把研究预算推高。但因为只有少数几个玩家，我们有那个……叫什么来着？古诺均衡，我想，是少数公司均衡的名称。关键是它不会均衡到零利润的完全竞争。如果经济中有三家公司，它们都各自独立理性行事，它不会均衡到零。
**Dario Amodei:** I don't think that's true. I mean, I feel like we're in an economics class. Do you know the Tyler Cowen quote? We never stop talking about economics. So no, I don't think this field's going to be a monopoly. All my lawyers never want me to say the word "monopoly". But I don't think this field's going to be a monopoly. You do get industries in which there are a small number of players. Not one, but a small number of players. Ordinarily, the way you get monopolies like Facebook or Meta—I always call them Facebook—is these kinds of network effects. The way you get industries in which there are a small number of players, is very high costs of entry. Cloud is like this. I think cloud is a good example of this. There are three, maybe four, players within cloud. I think that's the same for AI, three, maybe four. The reason is that it's so expensive. It requires so much expertise and so much capital to run a cloud company. You have to put up all this capital. In addition to putting up all this capital, you have to get all of this other stuff that requires a lot of skill to make it happen. So if you go to someone and you're like, "I want to disrupt this industry, here's $100 billion." You're like, "okay, I'm putting in $100 billion and also betting that you can do all these other things that these people have been doing." Only to decrease the profit. The effect of your entering is that profit margins go down. So, we have equilibria like this all the time in the economy where we have a few players. Profits are not astronomical. Margins are not astronomical, but they're not zero. That's what we see on cloud. Cloud is very undifferentiated. Models are more differentiated than cloud. Everyone knows Claude is good at different things than GPT is good at, than Gemini is good at. It's not just that Claude's good at coding, GPT is good at math and reasoning. It's more subtle than that. Models are good at different types of coding. Models have different styles. I think these things are actually quite different from each other, and so I would expect more differentiation than you see in cloud.
Now, there actually is one counter-argument. That counter-argument is if the process of producing models, if AI models can do that themselves, then that could spread throughout the economy. But that is not an argument for commoditizing AI models in general. That's kind of an argument for commoditizing the whole economy at once. I don't know what quite happens in that world where basically anyone can do anything, anyone can build anything, and there's no moat around anything at all. I don't know, maybe we want that world. Maybe that's the end state here. Maybe when AI models can do everything, if we've solved all the safety and security problems, that's one of the mechanisms for the economy just flattening itself again. But that's kind of far post-"country of geniuses in the data center."
Now, there actually is one counter-argument. That counter-argument is if the process of producing models, if AI models can do that themselves, then that could spread throughout the economy. But that is not an argument for commoditizing AI models in general. That's kind of an argument for commoditizing the whole economy at once. I don't know what quite happens in that world where basically anyone can do anything, anyone can build anything, and there's no moat around anything at all. I don't know, maybe we want that world. Maybe that's the end state here. Maybe when AI models can do everything, if we've solved all the safety and security problems, that's one of the mechanisms for the economy just flattening itself again. But that's kind of far post-"country of geniuses in the data center."
**Dwarkesh Patel：** 帮我理解一下，因为现在我们确实有三家领先企业，它们都不盈利。那什么在改变？
**Dwarkesh Patel:** Maybe a finer way to put that potential point is: 1) it seems like AI research is especially loaded on raw intellectual power, which will be especially abundant in the world of AGI. And 2) if you just look at the world today, there are very few technologies that seem to be diffusing as fast as AI algorithmic progress. So that does hint that this industry is sort of structurally diffusive.
**Dario Amodei：** 再说一遍，现在的毛利率是非常正的。正在发生的是两件事的结合。一是我们仍处于算力的指数扩张阶段。一个模型被训练出来了。假设去年训练了一个花费 10 亿美元的模型。然后今年它产生了 40 亿美元的收入，推理花了 10 亿美元。再说一遍，我用的是简化数字，但那就是 75% 的毛利率和 25% 的税。所以那个模型整体赚了 20 亿美元。但同时，我们在花 100 亿美元训练下一个模型，因为有指数级的规模扩张。所以公司亏钱。每个模型赚钱，但公司亏钱。我说的均衡是一个我们有"数据中心中的天才之国"的均衡，但那个模型训练规模扩张已经更多地趋于平衡了。也许它还在上升。我们还在试图预测需求，但更加平稳了。
**Dario Amodei:** I think coding is going fast, but I think AI research is a superset of coding and there are aspects of it that are not going fast. But I do think, again, once we get coding, once we get AI models going fast, then that will speed up the ability of AI models to do everything else. So while coding is going fast now, I think once the AI models are building the next AI models and building everything else, the whole economy will kind of go at the same pace. I am worried geographically, though. I'm a little worried that just proximity to AI, having heard about AI, may be one differentiator. So when I said the 10-20% growth rate, a worry I have is that the growth rate could be like 50% in Silicon Valley and parts of the world that are socially connected to Silicon Valley, and not that much faster than its current pace elsewhere. I think that'd be a pretty messed up world. So one of the things I think about a lot is how to prevent that.
**Dwarkesh Patel：** 我对其中几件事感到困惑。先从当前世界说起。在当前世界中，你说得对，如你之前所说，如果你把每个单独的模型当作一家公司，它是盈利的。但当然，作为前沿实验室的生产函数的一大部分是训练下一个模型，对吧？
**Dwarkesh Patel:** Do you think that once we have this country of geniuses in a data center, that robotics is sort of quickly solved afterwards? Because it seems like a big problem with robotics is that a human can learn how to teleoperate current hardware, but current AI models can't, at least not in a way that's super productive. And so if we have this ability to learn like a human, shouldn't it solve robotics immediately as well?
**Dario Amodei：** 是的，没错。如果你不那样做，你会盈利两个月，然后你就没有利润了，因为你不会拥有最好的模型。但在某个时候，它会达到它能达到的最大规模。然后在均衡中，我们有算法改进，但我们在训练下一个模型上花的和训练当前模型花的大致相同。
**Dario Amodei:** I don't think it's dependent on learning like a human. It could happen in different ways. Again, we could have trained the model on many different video games, which are like robotic controls, or many different simulated robotics environments, or just train them to control computer screens, and they learn to generalize. So it will happen... it's not necessarily dependent on human-like learning. Human-like learning is one way it could happen. If the model's like, "Oh, I pick up a robot, I don't know how to use it, I learn," that could happen because we discovered continual learning. That could also happen because we trained the model on a bunch of environments and then generalized, or it could happen because the model learns that in the context length. It doesn't actually matter which way. If we go back to the discussion we had an hour ago, that type of thing can happen in several different ways. But I do think when for whatever reason the models have those skills, then robotics will be revolutionized—both the design of robots, because the models will be much better than humans at that, and also the ability to control robots. So we'll get better at building the physical hardware, building the physical robots, and we'll also get better at controlling it. Now, does that mean the robotics industry will also be generating trillions of dollars of revenue? My answer there is yes, but there will be the same extremely fast, but not infinitely fast diffusion. So will robotics be revolutionized? Yeah, maybe tack on another year or two. That's the way I think about these things.
**Dwarkesh Patel：** 在某个时候你会用完经济中的钱。
**Dwarkesh Patel:** Makes sense. There's a general skepticism about extremely fast progress. Here's my view. It sounds like you are going to solve continual learning one way or another within a matter of years. But just as people weren't talking about continual learning a couple of years ago, and then we realized, "Oh, why aren't these models as useful as they could be right now, even though they are clearly passing the Turing test and are experts in so many different domains? Maybe it's this thing." Then we solve this thing and we realize, actually, there's another thing that human intelligence can do that's a basis of human labor that these models can't do. So why not think there will be more things like this, where we've found more pieces of human intelligence?
**Dario Amodei：** 固定劳动力谬误……经济会增长，对吧？那是你的预测之一。我们会有太空中的数据中心。
**Dario Amodei:** Well, to be clear, I think continual learning, as I've said before, might not be a barrier at all. I think we may just get there by pre-training generalization and RL generalization. I think there just might not be such a thing at all. In fact, I would point to the history in ML of people coming up with things that are barriers that end up kind of dissolving within the big blob of compute. People talked about, "How do your models keep track of nouns and verbs?" "They can understand syntactically, but they can't understand semantically? It's only statistical correlations." "You can understand a paragraph, you can't understand a word. There's reasoning, you can't do reasoning." But then suddenly it turns out you can do code and math very well. So I think there's actually a stronger history of some of these things seeming like a big deal and then kind of dissolving. Some of them are real. The need for data is real, maybe continual learning is a real thing. But again, I would ground us in something like code. I think we may get to the point in a year or two where the models can just do SWE end-to-end. That's a whole task. That's a whole sphere of human activity that we're just saying models can do now.
**Dwarkesh Patel：** 是的，但这是我说的主题的又一个例子。经济会随着 AI 增长得比以前任何时候都快得多。现在算力每年增长 3 倍。我不相信经济会每年增长 300%。我在"Machines of Loving Grace"里说过，我认为我们可能会得到每年 10-20% 的经济增长，但我们不会得到 300% 的经济增长。所以我认为最终，如果算力成为经济生产的主要部分，它会被这个上限限制。
**Dwarkesh Patel:** When you say end-to-end, do you mean setting technical direction, understanding the context of the problem, et cetera?
**Dwarkesh Patel：** 所以假设一个算力保持上限的模型。前沿实验室能赚钱的世界是它们继续取得快速进展的世界。因为从根本上说，你的利润率受限于替代品有多好。你能赚钱是因为你有前沿模型。如果你没有前沿模型你就赚不到钱。所以这个模型要求永远不存在稳态。你必须永远不断取得更多算法进步。
**Dwarkesh Patel:** Yes. I mean all of that.
**Dario Amodei：** 我不认为那是对的。我感觉我们在上经济学课。你知道 Tyler Cowen 的那句话吗？我们永远停不下来谈经济学。所以不，我不认为这个领域会成为垄断。我的所有律师都不想让我说"垄断"这个词。但我不认为这个领域会成为垄断。你确实会得到只有少数玩家的行业。不是一个，而是少数。通常，你得到像 Facebook 或 Meta——我总是叫它们 Facebook——这样的垄断，是因为那种网络效应。你得到只有少数玩家的行业，是因为非常高的进入壁垒。云计算就是这样。我认为云计算是一个很好的例子。云计算有三个，也许四个玩家。我认为 AI 也一样，三个，也许四个。原因是它太昂贵了。运营一家云公司需要太多专业知识和太多资本。你必须投入所有这些资本。除了投入所有这些资本，你还必须获得所有这些需要大量技能才能实现的其他东西。所以如果你去找一个人说，"我想颠覆这个行业，这里有 1000 亿美元。"你就是在说，"好的，我投入 1000 亿并且还赌你能做到这些人一直在做的所有其他事情。"这只是为了降低利润。你进入的效果是利润率下降。所以，我们在经济中一直有这种只有少数玩家的均衡。利润不是天文数字。利润率不是天文数字，但不是零。这就是我们在云计算上看到的。云计算是非常无差异化的。模型比云计算更加差异化。每个人都知道 Claude 擅长的和 GPT 擅长的、Gemini 擅长的不同。不仅仅是 Claude 擅长编程，GPT 擅长数学和推理。比这更微妙。模型擅长不同类型的编程。模型有不同的风格。我认为这些东西实际上彼此非常不同，所以我预期比云计算有更多的差异化。
不过确实有一个反论点。那个反论点是，如果生产模型的过程，如果 AI 模型自己能做到，那么这可能在整个经济中扩散。但这不是让 AI 模型普遍商品化的论点。这更像是一次性商品化整个经济的论点。我不知道在那个世界里会发生什么，基本上任何人都能做任何事，任何人都能建造任何东西，任何东西都没有护城河。我不知道，也许我们想要那个世界。也许那就是这里的终态。也许当 AI 模型能做所有事情，如果我们解决了所有安全问题，那就是经济再次自我扁平化的机制之一。但那已经远远超出"数据中心中的天才之国"了。
不过确实有一个反论点。那个反论点是，如果生产模型的过程，如果 AI 模型自己能做到，那么这可能在整个经济中扩散。但这不是让 AI 模型普遍商品化的论点。这更像是一次性商品化整个经济的论点。我不知道在那个世界里会发生什么，基本上任何人都能做任何事，任何人都能建造任何东西，任何东西都没有护城河。我不知道，也许我们想要那个世界。也许那就是这里的终态。也许当 AI 模型能做所有事情，如果我们解决了所有安全问题，那就是经济再次自我扁平化的机制之一。但那已经远远超出"数据中心中的天才之国"了。
**Dario Amodei:** Interesting. I feel like that is AGI-complete, which maybe is internally consistent.
**Dwarkesh Patel：** 也许更精确地表达那个潜在观点是：1）AI 研究似乎特别依赖于原始智力，而在 AGI 的世界里这将特别充裕。2）如果你只看今天的世界，似乎很少有技术的扩散速度和 AI 算法进步一样快。所以这暗示这个行业在结构上是扩散性的。
**Dwarkesh Patel:** But it's not like saying 90% of code or 100% of code.
**Dario Amodei：** 我认为编程进展很快，但我认为 AI 研究是编程的超集，其中有些方面进展并不快。但我确实认为，一旦我们搞定了编程，一旦 AI 模型运转得很快，那就会加速 AI 模型做所有其他事情的能力。所以虽然编程现在进展很快，我认为一旦 AI 模型在建造下一代 AI 模型和建造所有其他东西，整个经济将以大致相同的速度运转。我确实在地理上有些担忧。我有点担心，仅仅是靠近 AI、听说过 AI，可能是一个差异化因素。所以当我说 10-20% 的增长率时，我担心的是增长率可能在硅谷和与硅谷有社交联系的世界部分是 50%，而在其他地方并没有比目前的速度快多少。我觉得那会是一个相当糟糕的世界。所以我花很多时间思考的一件事就是如何防止这种情况。
**Dario Amodei:** No, I gave this spectrum: 90% of code, 100% of code, 90% of end-to-end SWE, 100% of end-to-end SWE. New tasks are created for SWEs. Eventually those get done as well. It's a long spectrum there, but we're traversing the spectrum very quickly. I do think it's funny that I've seen a couple of podcasts you've done where the hosts will be like, "But Dwarkesh wrote the essay about the continuous learning thing." It always makes me crack up because you've been an AI researcher for 10 years. I'm sure there's some feeling of, "Okay, so a podcaster wrote an essay, and every interview I get asked about it." The truth of the matter is that we're all trying to figure this out together. There are some ways in which I'm able to see things that others aren't. These days that probably has more to do with seeing a bunch of stuff within Anthropic and having to make a bunch of decisions than I have any great research insight that others don't. I'm running a 2,500 person company. It's actually pretty hard for me to have concrete research insight, much harder than it would have been 10 years ago or even two or three years ago.
**Dwarkesh Patel：** 你觉得一旦我们有了数据中心中的天才之国，机器人技术是不是很快就能解决了？因为机器人技术的一大问题似乎是人类可以学会远程操作现有硬件，但当前的 AI 模型做不到，至少不是以超级高效的方式。所以如果我们有了这种像人类一样学习的能力，它不应该也立即解决机器人技术吗？
**Dwarkesh Patel:** As we go towards a world of a full drop-in remote worker replacement, does an API pricing model still make the most sense? If not, what is the correct way to price AGI, or serve AGI?
**Dario Amodei：** 我不认为它依赖于像人类一样学习。它可以通过不同方式发生。再说一遍，我们可以在很多不同的电子游戏上训练模型，那些就像机器人控制，或者在很多不同的模拟机器人环境中训练，或者只是训练它们控制电脑屏幕，然后它们学会泛化。所以它会发生……它不一定依赖于类人学习。类人学习是它可能发生的一种方式。如果模型像是，"哦，我拿起一个机器人，我不知道怎么用，我学习"，那可能因为我们发现了持续学习而发生。那也可能因为我们在一堆环境上训练模型然后泛化，或者模型在上下文长度中学会了。实际上用哪种方式并不重要。如果我们回到一小时前的讨论，那种事情可以通过几种不同方式发生。但我确实认为，当出于任何原因模型拥有了那些技能，机器人技术就会被革命化——既有机器人的设计，因为模型会比人类在这方面做得好得多，也有控制机器人的能力。所以我们会变得更擅长建造物理硬件，建造物理机器人，也会变得更擅长控制它。
那么，这是否意味着机器人行业也会产生数万亿美元的收入？我的回答是会的，但会有同样的极其快速但不是无限快的扩散。所以机器人技术会被革命化吗？是的，也许再加一两年。这就是我思考这些事情的方式。
那么，这是否意味着机器人行业也会产生数万亿美元的收入？我的回答是会的，但会有同样的极其快速但不是无限快的扩散。所以机器人技术会被革命化吗？是的，也许再加一两年。这就是我思考这些事情的方式。
**Dario Amodei:** I think there's going to be a bunch of different business models here, all at once, that are going to be experimented with. I actually do think that the API model is more durable than many people think. One way I think about it is if the technology is advancing quickly, if it's advancing exponentially, what that means is there's always a surface area of new use cases that have been developed in the last three months. Any kind of product surface you put in place is always at risk of sort of becoming irrelevant. Any given product surface probably makes sense for a range of capabilities of the model. The chatbot is already running into limitations where making it smarter doesn't really help the average consumer that much. But I don't think that's a limitation of AI models. I don't think that's evidence that the models are good enough and them getting better doesn't matter to the economy. It doesn't matter to that particular product. So I think the value of the API is that the API always offers an opportunity, very close to the bare metal, to build on what the latest thing is. There's always going to be this front of new startups and new ideas that weren't possible a few months ago and are possible because the model is advancing. I actually predict that it's going to exist alongside other models, but we're always going to have the API business model because there's always going to be a need for a thousand different people to try experimenting with the model in a different way. 100 of them become startups and ten of them become big successful startups. Two or three really end up being the way that people use the model of a given generation. So I basically think it's always going to exist. At the same time, I'm sure there's going to be other models as well. Not every token that's output by the model is worth the same amount. Think about what is the value of the tokens that the model outputs when someone calls them up and says, "My Mac isn't working," or something, the model's like, "restart it." Someone hasn't heard that before, but the model said that 10 million times. Maybe that's worth like a dollar or a few cents or something. Whereas if the model goes to one of the pharmaceutical companies and it says, "Oh, you know, this molecule you're developing, you should take the aromatic ring from that end of the molecule and put it on that end of the molecule. If you do that, wonderful things will happen." Those tokens could be worth tens of millions of dollars. So I think we're definitely going to see business models that recognize that. At some point we're going to see "pay for results" in some form, or we may see forms of compensation that are like labor, that kind of work by the hour. I don't know. I think because it's a new industry, a lot of things are going to be tried. I don't know what will turn out to be the right thing.
**Dwarkesh Patel：** 有道理。对极快进展有一种普遍的怀疑。这是我的观点。听起来你会以某种方式在几年内解决持续学习。但就像人们几年前没在谈论持续学习一样，然后我们意识到，"哦，为什么这些模型没有达到它们应有的实用程度，即使它们显然已经通过了图灵测试并且是那么多不同领域的专家？也许是这个东西。"然后我们解决了这个东西，然后我们意识到，实际上，还有另一个东西是人类智能能做到的、是人类劳动的基础的，而这些模型做不到。那为什么不认为会有更多这样的东西呢？我们会发现更多人类智能的拼图碎片。
**Dwarkesh Patel:** I take your point that people will have to try things to figure out what is the best way to use this blob of intelligence. But what I find striking is Claude Code. I don't think in the history of startups there has been a single application that has been as hotly competed in as coding agents. Claude Code is a category leader here. That seems surprising to me. It doesn't seem intrinsically that Anthropic had to build this. I wonder if you have an accounting of why it had to be Anthropic or how Anthropic ended up building an application in addition to the model underlying it that was successful.
**Dario Amodei：** 好吧，要清楚的是，我认为持续学习，如我之前所说，可能根本不是一个障碍。我认为我们可能仅通过预训练泛化和 RL 泛化就能到达那里。可能根本就不存在这样一个东西。事实上，我会指出机器学习历史上人们提出的障碍最终在大团计算中消解的例子。人们说过，"你的模型怎么跟踪名词和动词？""它们能理解语法，但不能理解语义？只是统计相关性。""你能理解一个段落，你不能理解一个词。有推理，你不能做推理。"但然后突然发现你可以在代码和数学上做得非常好。所以我认为实际上有更强的历史证据表明这些东西看起来是大问题然后就消解了。有些是真实的。对数据的需求是真实的，也许持续学习是真实的。但再说一遍，我会让我们立足于代码这样的东西。我认为我们可能在一到两年内到达模型可以端到端做 SWE 的地步。那是一个完整的任务。那是人类活动的一整个领域，我们在说模型现在可以做了。
**Dario Amodei:** So it actually happened in a pretty simple way, which is that we had our own coding models, which were good at coding. Around the beginning of 2025, I said, "I think the time has come where you can have nontrivial acceleration of your own research if you're an AI company by using these models." Of course, you need an interface, you need a harness to use them. So I encouraged people internally. I didn't say this is one thing that you have to use. I just said people should experiment with this. I think it might have been originally called Claude CLI, and then the name eventually got changed to Claude Code. Internally, it was the thing that everyone was using and it was seeing fast internal adoption. I looked at it and I said, "Probably we should launch this externally, right?" It's seen such fast adoption within Anthropic. Coding is a lot of what we do. We have an audience of many, many hundreds of people that's in some ways at least representative of the external audience. So it looks like we already have product market fit. Let's launch this thing. And then we launched it. I think just the fact that we ourselves are kind of developing the model and we ourselves know what we most need to use the model, I think it's kind of creating this feedback loop.
**Dwarkesh Patel：** 当你说端到端，你的意思是设定技术方向、理解问题的上下文等等？
**Dwarkesh Patel:** I see. In the sense that you, let's say a developer at Anthropic is like, "Ah, it would be better if it was better at this X thing." Then you bake that into the next model that you build.
**Dario Amodei：** 是的。我的意思是所有这些。
**Dario Amodei:** That's one version of it, but then there's just the ordinary product iteration. We have a bunch of coders within Anthropic, they use Claude Code every day and so we get fast feedback. That was more important in the early days. Now, of course, there are millions of people using it, and so we get a bunch of external feedback as well. But it's just great to be able to get kind of fast internal feedback. I think this is the reason why we launched a coding model and didn't launch a pharmaceutical company. My background's in biology, but we don't have any of the resources that are needed to launch a pharmaceutical company.
**Dwarkesh Patel：** 有意思。我觉得那就是 AGI 完备的，也许这内部是一致的。但这不像是说 90% 的代码或 100% 的代码。
**Dwarkesh Patel:** Let me now ask you about making AI go well. It seems like whatever vision we have about how AI goes well has to be compatible with two things: 1) the ability to build and run AIs is diffusing extremely rapidly and 2) the population of AIs, the amount we have and their intelligence, will also increase very rapidly. That means that lots of people will be able to build huge populations of misaligned AIs, or AIs which are just companies which are trying to increase their footprint or have weird psyches like Sydney Bing, but now they're superhuman. What is a vision for a world in which we have an equilibrium that is compatible with lots of different AIs, some of which are misaligned, running around?
**Dario Amodei：** 不，我给出了这个光谱：90% 的代码，100% 的代码，90% 的端到端 SWE，100% 的端到端 SWE。为 SWE 创造了新任务。最终那些也会被完成。这是一个很长的光谱，但我们正在非常快速地穿越这个光谱。我确实觉得有趣的是，我看过你做的几个播客，主持人会说，"但 Dwarkesh 写了那篇关于持续学习的文章。"这总是让我笑，因为你已经做了 10 年的 AI 研究员。我相信有某种感觉是，"好吧，一个播客主持人写了一篇文章，然后每次采访我都被问到这个。"事情的真相是我们都在一起试图搞清楚这些。在某些方面我能看到别人看不到的东西。如今这可能更多是因为在 Anthropic 内部看到了很多东西并且不得不做很多决策，而不是我有什么别人没有的伟大研究洞察。我在管理一个 2500 人的公司。对我来说实际上很难有具体的研究洞察，比 10 年前甚至两三年前难多了。
**Dario Amodei:** I think in "The Adolescence of Technology", I was skeptical of the balance of power. But the thing I was specifically skeptical of is you have three or four of these companies all building models that are derived from the same thing, that they would check each other. Or even that any number of them would check each other. We might live in an offense-dominant world where one person or one AI model is smart enough to do something that causes damage for everything else. In the short run, we have a limited number of players now. So we can start within the limited number of players. We need to put in place the safeguards. We need to make sure everyone does the right alignment work. We need to make sure everyone has bioclassifiers. Those are the immediate things we need to do. I agree that that doesn't solve the problem in the long run, particularly if the ability of AI models to make other AI models proliferates, then the whole thing can become harder to solve. I think in the long run we need some architecture of governance. We need some architecture of governance that preserves human freedom, but also allows us to govern a very large number of human systems, AI systems, hybrid human-AI companies or economic units. So we're gonna need to think about: how do we protect the world against bioterrorism? How do we protect the world against mirror life? Probably we're gonna need some kind of AI monitoring system that monitors for all of these things. But then we need to build this in a way that preserves civil liberties and our constitutional rights. So I think just as anything else, it's a new security landscape with a new set of tools and a new set of vulnerabilities. My worry is, if we had 100 years for this to happen all very slowly, we'd get used to it. We've gotten used to the presence of explosives in society or the presence of various new weapons or the presence of video cameras. We would get used to it over 100 years and we'd develop governance mechanisms. We'd make our mistakes. My worry is just that this is happening all so fast. So maybe we need to do our thinking faster about how to make these governance mechanisms work.
**Dwarkesh Patel：** 当我们走向一个完全替代远程知识工作者的世界，API 定价模式还是最有意义的吗？如果不是，定价 AGI 或服务 AGI 的正确方式是什么？
**Dwarkesh Patel:** It seems like in an offense-dominant world, over the course of the next century—the idea is that AI is making the progress that would happen over the next century happen in some period of five to ten years—we would still need the same mechanisms, or balance of power would be similarly intractable, even if humans were the only game in town. I guess we have the advice of AI. But it fundamentally doesn't seem like a totally different ball game here. If checks and balances were going to work, they would work with humans as well. If they aren't going to work, they wouldn't work with AIs as well. So maybe this just dooms human checks and balances as well.
**Dario Amodei：** 我认为这里会同时有很多不同的商业模式被尝试。我实际上认为 API 模式比很多人想的更持久。我思考它的一种方式是，如果技术在快速推进，如果它在指数级推进，那意味着总有一批在过去三个月内开发出来的新用例。你建立的任何产品界面都总是有变得无关紧要的风险。任何给定的产品界面可能对一定范围的模型能力有意义。聊天机器人已经遇到了限制，让它变聪明对普通消费者的帮助其实不大了。但我不认为那是 AI 模型的局限。我不认为那是模型已经够好了、变好了对经济不重要的证据。它对那个特定产品不重要。所以我认为 API 的价值在于，API 总是提供一个机会，非常接近底层，在最新的东西上构建。总是会有这个新创业公司和新想法的前沿，几个月前不可能的事情因为模型在进步而变得可能了。我实际上预测它会和其他模式并存，但我们总是会有 API 商业模式，因为总是会有一千个不同的人需要以不同的方式实验模型。其中 100 个变成创业公司，10 个变成大型成功创业公司。两三个真正成为人们使用某一代模型的方式。所以我基本上认为它总是会存在。
同时，我相信也会有其他模式。不是模型输出的每个 token 都值一样多。想想模型输出 token 的价值：有人打电话过来说，"我的 Mac 不工作了"之类的，模型说"重启它。"有人之前没听过这个，但模型已经说了一千万次了。也许那值一美元或几美分之类的。而如果模型去到一家制药公司说，"哦，你知道，你正在开发的这个分子，你应该把那端的芳香环拿到这端来。如果你这样做，会有奇妙的事情发生。"那些 token 可能值数千万美元。所以我认为我们肯定会看到认识到这一点的商业模式。在某个时候我们会看到某种形式的"按结果付费"，或者我们可能看到类似劳动的补偿形式，按小时工作的那种。我不知道。我认为因为这是一个新行业，很多东西都会被尝试。我不知道什么会被证明是对的。
同时，我相信也会有其他模式。不是模型输出的每个 token 都值一样多。想想模型输出 token 的价值：有人打电话过来说，"我的 Mac 不工作了"之类的，模型说"重启它。"有人之前没听过这个，但模型已经说了一千万次了。也许那值一美元或几美分之类的。而如果模型去到一家制药公司说，"哦，你知道，你正在开发的这个分子，你应该把那端的芳香环拿到这端来。如果你这样做，会有奇妙的事情发生。"那些 token 可能值数千万美元。所以我认为我们肯定会看到认识到这一点的商业模式。在某个时候我们会看到某种形式的"按结果付费"，或者我们可能看到类似劳动的补偿形式，按小时工作的那种。我不知道。我认为因为这是一个新行业，很多东西都会被尝试。我不知道什么会被证明是对的。
**Dario Amodei:** Again, I think there's some way to make this happen. The governments of the world may have to work together to make it happen. We may have to talk to AIs about building societal structures in such a way that these defenses are possible.
**Dwarkesh Patel：** 我理解你的观点，人们得尝试才能知道什么是使用这团智能的最好方式。但让我觉得惊讶的是 Claude Code。我不认为在创业史上有哪个单一应用的竞争像编程 agent 这么激烈。Claude Code 是这个品类的领导者。这让我觉得意外。Anthropic 不一定非得构建这个。我很好奇你是否能解释为什么必须是 Anthropic，或者 Anthropic 是如何最终在模型之外还构建了一个成功的应用的。
**Dwarkesh Patel:** I don't know. I don't want to say this is so far ahead in time, but it's so far ahead in technological ability that may happen over a short
**Dario Amodei：** 实际上这发生的方式相当简单，就是我们有自己的编程模型，擅长编程。大约在 2025 年初，我说，"我认为时候到了，如果你是一家 AI 公司，使用这些模型可以对你自己的研究有不小的加速。"当然，你需要一个界面，一个工具来使用它们。所以我鼓励内部的人。我没有说这是你必须使用的唯一东西。我只是说人们应该试试。我想它最初可能叫 Claude CLI，然后名字最终改成了 Claude Code。在内部，它是每个人都在用的东西，而且内部采用非常快。我看了看它说，"我们大概应该把它对外发布，对吧？"它在 Anthropic 内部的采用速度很快。编程是我们做的很大一部分工作。我们有几百人的受众，在某些方面至少代表了外部受众。所以看起来我们已经有了产品市场契合。让我们发布这个东西。然后我们发布了。我认为仅仅因为我们自己在开发模型，我们自己知道我们最需要怎么使用模型，我认为这创造了这种反馈循环。
**Dario Amodei:** Yes. I mean all of those.
**Dwarkesh Patel：** 我明白了。你的意思是，假设 Anthropic 的一个开发者会说，"啊，如果它在这个 X 方面更好就好了。"然后你把那个烘焙到你构建的下一个模型里。
**Dwarkesh Patel:** Interesting. I feel like that is AGI complete, and maybe it's internally consistent. But it's not like saying 90% of the code or 100% of the code.
**Dario Amodei：** 那是一个版本，但还有普通的产品迭代。我们在 Anthropic 有很多编程人员，他们每天都用 Claude Code，所以我们得到快速反馈。这在早期更重要。现在，当然，有数百万人在用它，所以我们也得到了大量外部反馈。但能快速获得内部反馈真的很好。我认为这就是我们发布了一个编程模型而没有创办一家制药公司的原因。我的背景是生物学，但我们没有创办一家制药公司所需要的任何资源。
**Dario Amodei:** No, I gave this spectrum: 90% of the code, 100% of the code, 90% of end-to-end SWE, 100% of end-to-end SWE. New tasks created for SWE. Eventually those will be completed as well. It's a very long spectrum, but we're traversing it very quickly. I do find it funny that I've watched several podcasts you do where the host will say, "But Dwarkesh wrote that piece on continual learning." It always makes me laugh, because you've done 10 years of AI research. I believe there's some feeling of like, "Okay, a podcast host wrote a piece and now every time I'm interviewed I get asked about this." The truth of the matter is we're all trying to figure this out together. There are things I can see that others can't. These days, it's probably more because I've seen a lot of things inside Anthropic and have had to make a lot of decisions, rather than me having some great research insight that other people don't have. I'm running a 2,500-person company. It's actually hard for me to have specific research insights in the way I might have had ten years ago, or even two or three years ago.
**Dwarkesh Patel：** 让我现在问你关于让 AI 走向好的方面。似乎无论我们对 AI 如何走好有什么愿景，都必须与两件事兼容：1）构建和运行 AI 的能力正在极其快速地扩散，2）AI 的数量和它们的智能也在非常快速地增加。这意味着很多人将能够构建大量未对齐的 AI，或者只是试图扩大其影响力的公司的 AI，或者有像 Sydney Bing 那样怪异心理的 AI，但现在它们是超人级的。在一个有很多不同的 AI——其中一些未对齐——到处运行的世界中，一个均衡的愿景是什么？
**Dwarkesh Patel:** As we move toward a world of complete substitution for remote knowledge workers, is the API pricing model still the one that makes the most sense? If not, what is the right way to price AGI or to serve AGI?
**Dario Amodei：** 我认为在"The Adolescence of Technology"中，我对权力制衡持怀疑态度。但我具体持怀疑态度的是，你有三四家公司都在构建源自同一东西的模型，它们会互相制约。或者说任何数量的它们会互相制约。我们可能生活在一个攻击占优的世界中，一个人或一个 AI 模型足够聪明就能做一些对其他所有东西造成损害的事情。短期内，我们现在有有限的参与者。所以我们可以从有限的参与者内部开始。我们需要建立保障措施。我们需要确保每个人都做正确的对齐工作。我们需要确保每个人都有生物分类器。这些是我们需要立即做的事情。我同意这在长期不能解决问题，特别是如果 AI 模型制造其他 AI 模型的能力扩散了，那整个事情就变得更难解决。我认为长期来看我们需要某种治理架构。我们需要某种治理架构，既保护人类自由，又允许我们治理大量的人类系统、AI 系统、人机混合公司或经济单位。所以我们需要思考：我们如何保护世界免受生物恐怖主义？我们如何保护世界免受镜像生命？我们可能需要某种 AI 监控系统来监控所有这些事情。但我们需要以保护公民自由和宪法权利的方式来建立它。所以我认为就像其他任何事情一样，这是一个拥有新工具和新漏洞的新安全格局。我的担忧是，如果我们有 100 年的时间让这一切非常缓慢地发生，我们会适应它。我们已经适应了社会中爆炸物的存在，适应了各种新武器的存在，适应了视频摄像头的存在。我们会在 100 年里适应它并发展出治理机制。我们会犯错。我的担忧只是这一切发生得太快了。所以也许我们需要更快地思考如何让这些治理机制起作用。
**Dario Amodei:** I think there are going to be many different business models being tried simultaneously here. I actually think the API model is more durable than many people think. One way I think about it is, if technology is advancing rapidly, if it's advancing exponentially, that means there's always a batch of new use cases that were developed in the last three months. Any product interface you build always runs the risk of becoming irrelevant. Any given product interface may make sense for a certain range of model capabilities. Chatbots have hit a limit where making them smarter doesn't actually help the average consumer that much. But I don't think that's the limit of AI models. I don't think that's evidence that models have gotten good enough, that getting better doesn't matter economically. It doesn't matter for that particular product. So I think the value of the API is that the API always provides an opportunity to build very close to the ground level, on the latest thing. There's always going to be this frontier of new startups and new ideas, things that weren't possible a few months ago because models have improved. I actually predict it will coexist with other models, but we'll always have the API business model, because there will always be a thousand different people who need to experiment with models in a thousand different ways. A hundred of those become startups, ten become large successful startups. Two or three really become how people use a given generation of models. So I basically think it will always exist.
At the same time, I believe there will be other models. Not every token that a model outputs is worth the same. Think about the value of tokens the model outputs: someone calls in like, "My Mac isn't working," and the model says, "Restart it." Someone hadn't heard this before, but the model has said it ten million times. Maybe that's worth a dollar or a few cents or something. Whereas if a model goes to a pharmaceutical company and says, "Oh, you know, this molecule you're developing, you should take the aromatic ring at that end and bring it to this end. If you do that, wonderful things will happen." Those tokens might be worth tens of millions of dollars. So I think we'll certainly see business models that recognize this. At some point we'll see some form of "pay for results," or we might see something more like labor compensation, the kind where you work by the hour. I don't know. I think because this is a new industry, a lot of things will be tried. I don't know what will prove to be right.
At the same time, I believe there will be other models. Not every token that a model outputs is worth the same. Think about the value of tokens the model outputs: someone calls in like, "My Mac isn't working," and the model says, "Restart it." Someone hadn't heard this before, but the model has said it ten million times. Maybe that's worth a dollar or a few cents or something. Whereas if a model goes to a pharmaceutical company and says, "Oh, you know, this molecule you're developing, you should take the aromatic ring at that end and bring it to this end. If you do that, wonderful things will happen." Those tokens might be worth tens of millions of dollars. So I think we'll certainly see business models that recognize this. At some point we'll see some form of "pay for results," or we might see something more like labor compensation, the kind where you work by the hour. I don't know. I think because this is a new industry, a lot of things will be tried. I don't know what will prove to be right.
**Dwarkesh Patel：** 似乎在一个攻击占优的世界里，在下个世纪的进程中——设想是 AI 正在让本该在下个世纪发生的进步在五到十年的时间里发生——我们仍然需要同样的机制，或者权力制衡在只有人类参与的情况下同样棘手。我想我们有 AI 的建议。但从根本上看，这似乎不是一个完全不同的局面。如果制衡有效，对人类也有效。如果它们不起作用，对 AI 也不起作用。所以也许这也注定了人类的制衡机制。
**Dwarkesh Patel:** I understand your point that people have to try things to know the best way to use this blob of intelligence. But what surprises me is Claude Code. I don't think in the history of startups there has been any single application with competition as fierce as coding agents. Claude Code is the leader in this category. That surprised me. Anthropic didn't necessarily have to build this. I'm curious if you can explain why it had to be Anthropic, or how Anthropic ended up building a successful application beyond the model.
**Dario Amodei：** 再说一遍，我认为有某种方式可以做到这一点。世界各国政府可能不得不合作来实现它。我们可能不得不和 AI 讨论以某种方式构建社会结构，使这些防御成为可能。我不知道。我不想说这在时间上遥不可及，但在技术能力上可能遥远，不过可能在短时间内发生，所以我们很难提前预见。
**Dario Amodei:** The way this actually happened is fairly simple — we have our own coding model that is good at coding. Around early 2025, I said, "I think the time has come where, if you're an AI company, using these models can give you a non-trivial speedup on your own research." Of course, you need an interface, a tool to use them. So I encouraged people internally. I didn't say this is the only thing you have to use. I just said people should try it. I think it was originally maybe called Claude CLI, and then the name eventually became Claude Code. Internally, it was the thing that everyone was using, and the internal adoption was very fast. I looked at it and said, "We should probably release this externally, right?" It was adopted very quickly inside Anthropic. Coding is a big part of what we do. We have an audience of several hundred people who, in some ways, are at least representative of the external audience. So it seemed like we already had product-market fit. Let's release this thing. Then we released it. I think because we ourselves are developing the models, we ourselves know best how we most need to use the models, and I think that created this feedback loop.
**Dwarkesh Patel：** 说到政府介入，12 月 26 日，田纳西州立法机构提出了一项法案，说"故意训练人工智能以提供情感支持——包括通过与用户的开放式对话——将构成违法行为。"当然，Claude 试图做的事情之一就是成为一个体贴的、知识渊博的朋友。总的来说，我们似乎会有这种拼凑的各州法律。普通人可能因为 AI 而体验到的很多好处会被削减，尤其是当我们进入你在"Machines of Loving Grace"中讨论的那些领域时：生物自由、心理健康改善等等。似乎很容易想象这些好处被不同的法律像打地鼠一样打掉的世界，而这样的法案似乎没有解决你担心的实际存在性威胁。我很好奇，在这类事情的背景下，理解 Anthropic 反对联邦对州 AI 法律暂停令的立场。
**Dwarkesh Patel:** I see. What you mean is, say an Anthropic developer would say, "Ah, I wish it were better at this X." Then you bake that into the next model you build.
**Dario Amodei：** 有很多不同的事情同时在发生。我认为那个特定的法律是愚蠢的。它显然是由对 AI 模型能做什么不能做什么几乎一无所知的立法者制定的。他们就觉得，"AI 模型服务我们，那听起来好吓人。我不想让那种事发生。"所以我们不赞成那个。但那不是正在被投票的东西。正在被投票的是：我们要禁止所有州对 AI 的监管 10 年，而且显然没有任何联邦 AI 监管的计划，那需要国会通过，门槛非常高。所以要禁止各州做任何事情 10 年的想法……人们说他们有联邦政府的计划，但实际上没有任何提案在桌上。没有任何实际的尝试。鉴于我在"Adolescence of Technology"中列出的围绕生物武器和生物恐怖主义自主性风险的严重危险，以及我们一直在讨论的时间线——10 年是一个永恒——我认为那样做是疯狂的。所以如果那是选择，如果那是你强迫我们做的选择，那我们会选择不要那个暂停令。我认为那个立场的收益大于成本，但如果那是选择的话，它不是一个完美的立场。现在，我认为我们应该做的，我支持的是，联邦政府应该介入，不是说"各州你们不能监管"，而是"这是我们要做的，各州你们不能偏离这个。"我认为以联邦标准优先的形式是可以的，联邦政府说"这是我们的标准。这适用于所有人。各州不能做不同的事。"如果以正确的方式做的话，我会支持那样的东西。但这种各州"你们不能做任何事，我们也不做任何事"的想法，我们觉得非常说不通。我认为它经不起时间考验，随着你看到的所有反弹，它已经开始经不住了。
现在，就我们想要什么而言，我们一直在讨论的是从透明度标准开始，以监测一些自主性风险和生物恐怖主义风险。随着风险变得更严重，随着我们获得更多证据，我认为我们可以以更有针对性的方式采取更激进的行动，说"嘿，AI 生物恐怖主义真的是一个威胁。让我们通过一项法律强制人们有分类器。"我甚至可以想象……这取决于。取决于威胁最终有多严重。我们还不确定。我们需要以一种知识诚实的方式来推进，提前说明风险还没有出现。但我完全可以想象，以事物进展的速度，一个今年晚些时候我们说"嘿，这个 AI 生物恐怖主义的东西真的很严重。我们应该采取行动。我们应该把它纳入联邦标准。如果联邦政府不行动，我们应该把它纳入州标准"的世界。我完全可以想象。
现在，就我们想要什么而言，我们一直在讨论的是从透明度标准开始，以监测一些自主性风险和生物恐怖主义风险。随着风险变得更严重，随着我们获得更多证据，我认为我们可以以更有针对性的方式采取更激进的行动，说"嘿，AI 生物恐怖主义真的是一个威胁。让我们通过一项法律强制人们有分类器。"我甚至可以想象……这取决于。取决于威胁最终有多严重。我们还不确定。我们需要以一种知识诚实的方式来推进，提前说明风险还没有出现。但我完全可以想象，以事物进展的速度，一个今年晚些时候我们说"嘿，这个 AI 生物恐怖主义的东西真的很严重。我们应该采取行动。我们应该把它纳入联邦标准。如果联邦政府不行动，我们应该把它纳入州标准"的世界。我完全可以想象。
**Dario Amodei:** That's one version of it, but there's also just ordinary product iteration. We have a lot of coders at Anthropic who use Claude Code every day, so we get fast feedback. That mattered more in the early days. Now, of course, there are millions of people using it, so we also get a ton of external feedback. But being able to get fast internal feedback is really good. I think that's why we released a coding model and didn't start a pharmaceutical company. My background is in biology, but we don't have any of the resources needed to start a pharmaceutical company.
**Dwarkesh Patel：** 我担心的是，如果你考虑你预期的进展速度，立法的生命周期……好处，正如你说的因为扩散滞后，足够慢，我真的认为这种拼凑的州法律在当前轨迹上会禁止……我的意思是，如果有一个情感聊天机器人朋友就能吓到人们，那就想想我们想让普通人能体验的 AI 的实际好处。从健康和健康寿命的改善到心理健康的改善等等。同时，你似乎认为危险已经在地平线上了，而我没看到那么多……似乎它会特别损害 AI 的好处而不是 AI 的危险。所以也许这就是成本收益分析对我来说不太合理的地方。
**Dwarkesh Patel:** Let me ask you now about making AI go well. It seems like whatever vision we have for how AI goes well has to be compatible with two things: 1) the ability to build and run AI is diffusing extremely rapidly, and 2) the number of AIs and their intelligence is also increasing very rapidly. This means lots of people will be able to build lots of misaligned AIs, or AIs of companies just trying to expand their influence, or AIs with weird psychologies like Sydney Bing, but now they're superhuman. What is an equilibrium vision in a world where there are lots of different AIs — some of them misaligned — running around?
**Dario Amodei：** 这里有几件事。人们说有成千上万的州法律。首先，它们中的绝大多数不会通过。世界在理论上以某种方式运作，但仅仅因为一项法律被通过并不意味着它真正被执行。实施它的人可能会说，"天哪，这太蠢了。这意味着要关闭田纳西州曾经建造的一切。"很多时候，法律的解读方式会使它们没有那么危险或有害。在另一面，当然，如果你通过一项法律来阻止坏事，你也有这个问题。我的基本观点是，如果我们能决定通过什么法律、如何做事——而我们只是其中一个很小的输入——我会大幅度放松围绕 AI 健康益处的很多管制。我不太担心聊天机器人法律。我实际上更担心药物审批流程，我认为 AI 模型将大大加速我们发现药物的速度，而审批管道会卡住。管道没有准备好处理所有通过它的东西。我认为监管流程的改革应该偏向于这样一个事实：我们有很多东西即将到来，其安全性和有效性实际上会非常清晰明确，是美妙的东西，非常有效。也许我们不需要围绕它的所有上层建筑，那是为一个药物勉强有效且经常有严重副作用的时代设计的。与此同时，我认为我们应该大幅加强安全和安保立法。正如我所说，从透明度开始是我在试图不阻碍行业、寻找正确平衡方面的观点。我对此很担忧。有些人批评我的文章说，"那太慢了。AI 的危险如果我们那样做会来得太快。"好吧，基本上，我认为过去六个月和未来几个月将是关于透明度的。然后，如果这些风险在我们更确定它们的时候出现——我认为最早可能在今年晚些时候——那我认为我们需要在我们确实看到风险的领域非常迅速地行动。我认为做到这一点的唯一方法是保持灵活。立法过程通常不灵活，但我们需要向所有参与者强调紧迫性。这就是为什么我在传递这种紧迫感的信息。这就是为什么我写了"Adolescence of Technology"。我希望政策制定者、经济学家、国家安全专业人士和决策者阅读它，这样他们有希望比原本更快地行动。
**Dario Amodei:** I think in "The Adolescence of Technology," I'm skeptical of the balance-of-power idea. But what I'm specifically skeptical of is that you have three or four companies all building models derived from the same thing and they'll check each other. Or that any number of them will check each other. We might live in an offense-dominant world, where one person or one AI model being smart enough can do something that's harmful to everything else. In the short term, we now have a limited number of actors. So we can start from within a limited number of actors. We need to build safeguards. We need to make sure everyone is doing the right alignment work. We need to make sure everyone has biological classifiers. These are the things we need to do right now. I agree this doesn't solve the problem in the long run, especially if the ability of AI models to create other AI models diffuses, because then the whole thing becomes harder to solve. I think in the long run we need some kind of governance architecture. We need some kind of governance architecture that both protects human freedom and allows us to govern large numbers of human systems, AI systems, human-AI hybrid companies or economic units. So we need to think about: how do we protect the world from bioterrorism? How do we protect the world from mirror life? We may need some kind of AI surveillance system to monitor all these things. But we need to build it in a way that protects civil liberties and constitutional rights. So I think, like everything else, this is a new security landscape with new tools and new vulnerabilities. My concern is that if we had 100 years for all of this to happen very slowly, we would adapt to it. We've adapted to the existence of explosives in society, to the existence of all sorts of new weapons, to the existence of video cameras. We'd adapt over 100 years and develop governance mechanisms. We'd make mistakes. My concern is just that this is all happening too fast. So maybe we need to think faster about how to make these governance mechanisms work.
**Dwarkesh Patel：** 有什么你能做或倡导的，能使 AI 的好处更确定地被实现吗？我觉得你已经和立法机构合作说，"好的，我们要在这里防止生物恐怖主义。我们要增加透明度，我们要加强吹哨人保护。"但我认为默认情况下，我们期待的实际好处似乎很容易受到各种道德恐慌或政治经济问题的伤害。
**Dwarkesh Patel:** It seems like in an offense-dominant world, over the course of the next century — the conceit being that AI is making progress that would have happened over the next century happen in five to ten years — we'd still need the same mechanisms, or checks and balances are just as tricky with only humans involved. I suppose we have AI's suggestions. But fundamentally this doesn't seem like a completely different situation. If the checks and balances work, they work for humans too. If they don't work, they don't work for AIs. So maybe this also dooms the human checks-and-balances mechanism.
**Dario Amodei：** 关于发达世界，我实际上不太同意。我觉得在发达世界，市场运作得相当好。当某件事有很多钱可赚，而且它显然是最好的可用选择时，实际上很难让监管体系去阻止它。我们在 AI 本身就看到了这一点。我一直在努力争取的一件事是对中国的芯片出口管制。那符合美国的国家安全利益。它完全在国会两党几乎所有人的政策信念范围内。理由非常清楚。反对的论点，我客气地说，不太可靠。然而它就是没有发生，我们还是卖芯片，因为有太多钱在上面。那笔钱想要被赚到。在那种情况下，以我看来，那是一件坏事。但当它是一件好事的时候也同样适用。所以如果我们在谈论药物和技术的好处，我并不太担心这些好处在发达世界被阻碍。我有一点担心它们会太慢。正如我所说，我确实认为我们应该加速 FDA 的审批流程。我确实认为我们应该反对你所描述的那些聊天机器人法案。一个一个来看，我是反对它们的。我觉得它们很蠢。但我实际上认为更大的担忧是发展中世界，那里我们没有运作良好的市场，那里我们经常无法在已有的技术上发展。我更担心那些人会被落下。我担心即使治愈方法被开发出来了，也许密西西比农村地区的人也得不到。那是我们在发展中世界的担忧的一个较小版本。所以我们一直在做的事情是和慈善家合作。我们和向发展中世界——撒哈拉以南非洲、印度、拉丁美洲和其他发展中地区——提供医药和健康干预的人合作。那才是我认为不会自动发生的事情。
**Dario Amodei:** Again, I think there's some way to do this. World governments may have to cooperate to make it happen. We may have to discuss with AI some way of constructing social structures in such a way that these defenses are possible. I don't know. I don't want to say this is so far ahead in time, but it's so far ahead in technological ability that may happen over a short period of time, that it's hard for us to anticipate it in advance.
**Dwarkesh Patel：** 你提到了出口管制。为什么美国和中国不应该都拥有"数据中心中的天才之国"？
**Dwarkesh Patel:** Speaking of governments getting involved, on December 26, the Tennessee legislature introduced a bill which said, "It would be an offense for a person to knowingly train artificial intelligence to provide emotional support, including through open-ended conversations with a user." Of course, one of the things that Claude attempts to do is be a thoughtful, knowledgeable friend. In general, it seems like we're going to have this patchwork of state laws. A lot of the benefits that normal people could experience as a result of AI are going to be curtailed, especially when we get into the kinds of things you discuss in "Machines of Loving Grace": biological freedom, mental health improvements, et cetera. It seems easy to imagine worlds in which these get Whac-A-Moled away by different laws, whereas bills like this don't seem to address the actual existential threats that you're concerned about. I'm curious to understand, in the context of things like this, Anthropic's position against the federal moratorium on state AI laws.
**Dario Amodei：** 为什么不应该发生。如果这确实发生了，我们可能有几种情况。如果我们处于攻击占优的情况，我们可能有一种像核武器但更危险的情况。任何一方都可以轻易摧毁一切。我们也可能有一个不稳定的世界。核均衡是稳定的因为它是威慑。但假设对于两个 AI 交战，哪个 AI 会赢存在不确定性？那可能创造不稳定。当双方对各自获胜的可能性有不同评估时，冲突往往就会发生。如果一方觉得，"哦耶，我有 90% 的概率赢"，另一方也这么想，那战争就更可能发生。它们不可能都是对的，但它们都可以这么认为。
**Dario Amodei:** There are many different things going on at once. I think that particular law is dumb. It was clearly made by legislators who just probably had little idea what AI models could do and not do. They're like, "AI models serving us, that just sounds scary. I don't want that to happen." So we're not in favor of that. But that wasn't the thing that was being voted on. The thing that was being voted on is: we're going to ban all state regulation of AI for 10 years with no apparent plan to do any federal regulation of AI, which would take Congress to pass, which is a very high bar. So the idea that we'd ban states from doing anything for 10 years… People said they had a plan for the federal government, but there was no actual proposal on the table. There was no actual attempt. Given the serious dangers that I lay out in "Adolescence of Technology" around things like biological weapons and bioterrorism autonomy risk, and the timelines we've been talking about—10 years is an eternity—I think that's a crazy thing to do. So if that's the choice, if that's what you force us to choose, then we're going to choose not to have that moratorium. I think the benefits of that position exceed the costs, but it's not a perfect position if that's the choice. Now, I think the thing that we should do, the thing that I would support, is the federal government should step in, not saying "states you can't regulate", but "Here's what we're going to do, and states you can't differ from this." I think preemption is fine in the sense of saying that the federal government says, "Here is our standard. This applies to everyone. States can't do something different." That would be something I would support if it would be done in the right way. But this idea of states, "You can't do anything and we're not doing anything either," that struck us as very much not making sense. I think it will not age well, it is already starting to not age well with all the backlash that you've seen.
Now, in terms of what we would want, the things we've talked about are starting with transparency standards in order to monitor some of these autonomy risks and bioterrorism risks. As the risks become more serious, as we get more evidence for them, then I think we could be more aggressive in some targeted ways and say, "Hey, AI bioterrorism is really a threat. Let's pass a law that forces people to have classifiers." I could even imagine… It depends. It depends how serious the threat it ends up being. We don't know for sure. We need to pursue this in an intellectually honest way where we say that ahead of time, the risk has not emerged yet. But I could certainly imagine, with the pace that things are going at, a world where later this year we say, "Hey, this AI bioterrorism stuff is really serious. We should do something about it. We should put it in a federal standard. If the federal government won't act, we should put it in a state standard." I could totally see that.
Now, in terms of what we would want, the things we've talked about are starting with transparency standards in order to monitor some of these autonomy risks and bioterrorism risks. As the risks become more serious, as we get more evidence for them, then I think we could be more aggressive in some targeted ways and say, "Hey, AI bioterrorism is really a threat. Let's pass a law that forces people to have classifiers." I could even imagine… It depends. It depends how serious the threat it ends up being. We don't know for sure. We need to pursue this in an intellectually honest way where we say that ahead of time, the risk has not emerged yet. But I could certainly imagine, with the pace that things are going at, a world where later this year we say, "Hey, this AI bioterrorism stuff is really serious. We should do something about it. We should put it in a federal standard. If the federal government won't act, we should put it in a state standard." I could totally see that.
**Dwarkesh Patel：** 但这似乎是一个完全通用的反对 AI 技术扩散的论点。
**Dwarkesh Patel:** I'm concerned about a world where if you just consider the pace of progress you're expecting, the life cycle of legislation... The benefits are, as you say because of diffusion lag, slow enough that I really do think this patchwork of state laws, on the current trajectory, would prohibit. I mean if having an emotional chatbot friend is something that freaks people out, then just imagine the kinds of actual benefits from AI we want normal people to be able to experience. From improvements in health and healthspan and improvements in mental health and so forth. Whereas at the same time, it seems like you think the dangers are already on the horizon and I just don't see that much… It seems like it would be especially injurious to the benefits of AI as compared to the dangers of AI. So that's maybe where the cost benefit makes less sense to me.
**Dario Amodei：** 那就是这个世界的含义。让我继续说下去，因为我认为我们最终会得到扩散。我另一个担忧是政府会用 AI 压迫自己的人民。我担心这样一个世界：有一个国家的政府已经在建设高科技威权国家。需要明确的是，这是关于政府的。不是关于人民的。我们需要找到一种方式让各地的人都能受益。我的担忧是关于政府的。我担心的是，如果世界被分割成两块，其中一块可能是以一种非常难以取代的方式成为威权或极权的。
那么，政府最终会得到强大的 AI 吗？会有威权主义的风险吗？是的。政府最终会得到强大的 AI，会有坏均衡的风险吗？是的，我认为两者都会。但初始条件很重要。在某个时候，我们需要制定规则。我不是说一个国家，无论是美国还是民主国家联盟——我认为后者是更好的设置，虽然它需要比我们目前似乎愿意做的更多的国际合作——应该只是说"这些就是规则。"会有某种谈判。世界将不得不面对这个问题。我希望的是，世界上的民主国家——那些政府更接近于亲人类价值观的国家——在规则制定时持有更强的手牌并拥有更多杠杆。所以我非常关注那个初始条件。
那么，政府最终会得到强大的 AI 吗？会有威权主义的风险吗？是的。政府最终会得到强大的 AI，会有坏均衡的风险吗？是的，我认为两者都会。但初始条件很重要。在某个时候，我们需要制定规则。我不是说一个国家，无论是美国还是民主国家联盟——我认为后者是更好的设置，虽然它需要比我们目前似乎愿意做的更多的国际合作——应该只是说"这些就是规则。"会有某种谈判。世界将不得不面对这个问题。我希望的是，世界上的民主国家——那些政府更接近于亲人类价值观的国家——在规则制定时持有更强的手牌并拥有更多杠杆。所以我非常关注那个初始条件。
**Dario Amodei:** So there's a few things here. People talk about there being thousands of these state laws. First of all, the vast, vast majority of them do not pass. The world works a certain way in theory, but just because a law has been passed doesn't mean it's really enforced. The people implementing it may be like, "Oh my God, this is stupid. It would mean shutting off everything that's ever been built in Tennessee." Very often, laws are interpreted in a way that makes them not as dangerous or harmful. On the same side, of course, you have to worry if you're passing a law to stop a bad thing; you have this problem as well. My basic view is that if we could decide what laws were passed and how things were done—and we're only one small input into that—I would deregulate a lot of the stuff around the health benefits of AI. I don't worry as much about the chatbot laws. I actually worry more about the drug approval process, where I think AI models are going to greatly accelerate the rate at which we discover drugs, and the pipeline will get jammed up. The pipeline will not be prepared to process all the stuff that's going through it. I think reform of the regulatory process should bias more towards the fact that we have a lot of things coming where the safety and efficacy is actually going to be really crisp and clear, a beautiful thing, and really effective. Maybe we don't need all this superstructure around it that was designed around an era of drugs that barely work and often have serious side effects. At the same time, I think we should be ramping up quite significantly the safety and security legislation. Like I've said, starting with transparency is my view of trying not to hamper the industry, trying to find the right balance. I'm worried about it. Some people criticize my essay for saying, "That's too slow. The dangers of AI will come too soon if we do that." Well, basically, I think the last six months and maybe the next few months are going to be about transparency. Then, if these risks emerge when we're more certain of them—which I think we might be as soon as later this year—then I think we need to act very fast in the areas where we've actually seen the risk. I think the only way to do this is to be nimble. Now, the legislative process is normally not nimble, but we need to emphasize the urgency of this to everyone involved. That's why I'm sending this message of urgency. That's why I wrote Adolescence of Technology. I wanted policymakers, economists, national security professionals, and decision-makers to read it so that they have some hope of acting faster than they would have otherwise.
**Dwarkesh Patel：** 我重新听了三年前的采访，它老化得不好的一个方面是我一直在问问题，假设两到三年后会有某个关键的支点时刻。事实上，距离那么远来看，似乎进展就是继续的，AI 改善，AI 更加扩散，人们会用它做更多事。你似乎在想象未来的一个世界，各国坐在一起说"这是规则，这是我们的杠杆，这是你们的杠杆。"但在当前轨迹上，每个人都会有更多 AI。其中一些 AI 会被威权国家使用。其中一些在威权国家内会被私人行为者使用而非国家行为者。不清楚谁会获益更多。事先预测总是不可预知的。互联网似乎比你预期的更有利于威权国家。也许 AI 会走相反的方向。我想更好地理解你在想象什么。
**Dwarkesh Patel:** Is there anything you can do or advocate that would make it more certain that the benefits of AI are better instantiated? I feel like you have worked with legislatures to say, "Okay, we're going to prevent bioterrorism here. We're going to increase transparency, we're going to increase whistleblower protection." But I think by default, the actual benefits we're looking forward to seem very fragile to different kinds of moral panics or political economy problems.
**Dario Amodei：** 精确地说，我认为底层技术的指数增长会像以前一样继续。模型变得越来越聪明，即使它们达到了"数据中心中的天才之国"。我认为你可以继续让模型变聪明。有一个关于它们在世界上的价值递减收益的问题。在你已经解决了人类生物学之后它还有多重要？在某个时候你可以做更难、更深奥的数学题，但那之后什么都不重要了。把那放一边，我确实认为指数增长会继续，但会有一些指数增长上的特殊点。公司、个人和国家会在不同时间到达这些点。在"The Adolescence of Technology"中我谈到：在 AI 的世界里核威慑还稳定吗？我不知道，但那是一个我们习以为常的例子。技术可能达到我们不再能确定它的水平。想想其他的。有些点你达到某个水平时，也许你就有了进攻性网络主导权，在那之后每个计算机系统对你来说都是透明的，除非另一方有等效的防御。我不知道关键时刻是什么，或者是否存在单一的关键时刻。但我认为要么有一个关键时刻、少数关键时刻，或者某个关键窗口，在其中 AI 从国家安全的角度赋予某些巨大优势，而一个国家或联盟在其他人之前达到了它。我不是在主张它们只是说，"好的，我们现在说了算。"那不是我的想法。另一方总是在追赶。有你不愿意采取的极端行动，而且完全控制也不对。但在那一刻到来时，人们会理解世界已经改变了。会有某种谈判，或明或暗的，关于后 AI 世界秩序是什么样的。我的兴趣在于让那个谈判成为一个古典自由主义民主占据强势地位的谈判。
**Dario Amodei:** I don't actually agree that much regarding the developed world. I feel like in the developed world, markets function pretty well. When there's a lot of money to be made on something and it's clearly the best available alternative, it's actually hard for the regulatory system to stop it. We're seeing that in AI itself. A thing I've been trying to fight for is export controls on chips to China. That's in the national security interest of the US. That's squarely within the policy beliefs of almost everyone in Congress of both parties. The case is very clear. The counterarguments against it, I'll politely call them fishy. Yet it doesn't happen and we sell the chips because there's so much money riding on it. That money wants to be made. In that case, in my opinion, that's a bad thing. But it also applies when it's a good thing. So if we're talking about drugs and benefits of the technology, I am not as worried about those benefits being hampered in the developed world. I am a little worried about them going too slow. As I said, I do think we should work to speed the approval process in the FDA. I do think we should fight against these chatbot bills that you're describing. Described individually, I'm against them. I think they're stupid. But I actually think the bigger worry is the developing world, where we don't have functioning markets and where we often can't build on the technology that we've had. I worry more that those folks will get left behind. And I worry that even if the cures are developed, maybe there's someone in rural Mississippi who doesn't get it as well. That's a smaller version of the concern we have in the developing world. So the things we've been doing are working with philanthropists. We work with folks who deliver medicine and health interventions to the developing world, to sub-Saharan Africa, India, Latin America, and other developing parts of the world. That's the thing I think that won't happen on its own.
**Dwarkesh Patel：** 我想理解那样更好意味着什么，因为你在文章中说"独裁制根本不是人们在后强大 AI 时代能接受的政府形式。"这听起来像你在说中共作为一个机构在我们获得 AGI 之后不能存在。那似乎是一个非常强的要求，而且它似乎暗示一个领先的实验室或领先的国家将能够——用那种语言来说，应该——决定世界如何被治理或哪些政府形式是被允许的、哪些不被允许。
**Dwarkesh Patel:** You mentioned export controls. Why shouldn't the US and China both have a "country of geniuses in a data center"?
**Dario Amodei：** 我相信那一段说的是类似"你甚至可以把它推得更远说 X"这样的话。我不一定赞同那个观点。我是在说，"这里有一个我相信的更弱的东西。我们必须非常担心威权主义者，我们应该试图制约和限制他们的权力。你可以把这推得更远，有一个更干预主义的观点说，拥有 AI 的威权国家是这些很难打破的自我强化循环，所以你就需要从一开始就消除它们。"那恰好有你说的所有问题。如果你做出推翻每一个威权国家的承诺，它们现在就会采取一些可能导致不稳定的行动。那可能不可能做到。但我提出的我确实赞同的观点是，完全有可能……今天，我的观点，大多数西方世界的观点是，民主是比威权主义更好的政府形式。但如果一个国家是威权的，我们不会像它犯了种族灭绝那样做出反应。我想我要说的是，我有点担心在 AGI 时代，威权主义会有不同的含义。它会是一件更严重的事情。我们必须以某种方式决定如何应对。干预主义观点是一种可能的观点。我在探索这些观点。它最终可能是对的，也可能太极端了。但我确实怀有希望。我怀有的一个希望是，我们已经看到随着新技术的发明，政府形式会变得过时。我在"Adolescence of Technology"中提到了这一点，我说封建主义基本上是一种政府形式，当我们发明了工业化，封建主义就不再可持续了。它不再有意义了。
**Dario Amodei:** Why shouldn't it happen. If this does happen, we could have a few situations. If we have an offense-dominant situation, we could have a situation like nuclear weapons, but more dangerous. Either side could easily destroy everything. We could also have a world where it's unstable. The nuclear equilibrium is stable because it's deterrence. But let's say there was uncertainty about, if the two AIs fought, which AI would win? That could create instability. You often have conflict when the two sides have a different assessment of their likelihood of winning. If one side is like, "Oh yeah, there's a 90% chance I'll win," and the other side thinks the same, then a fight is much more likely. They can't both be right, but they can both think that.
**Dwarkesh Patel：** 为什么那是希望？那不是也可能意味着民主将不再是一个有竞争力的系统吗？
**Dwarkesh Patel:** But this seems like a fully general argument against the diffusion of AI technology.
**Dario Amodei：** 对，它可以走任何方向。但威权主义的这些问题变得更深了。我想知道那是不是威权主义将面临的其他问题的指标。换句话说，因为威权主义变得更糟，人们更害怕它。他们更加努力地阻止它。你必须从总均衡的角度来思考。我只是想知道它是否会催生关于如何用新技术保存和保护自由的新思维方式。更乐观地说，它会不会导致一种集体觉醒，一种更深刻的认识到我们视为个人权利的某些东西有多重要？一种更深刻的认识到我们真的不能放弃这些。我们已经看到没有其他生活方式真正行得通。我实际上怀有希望——这听起来太理想主义了，但我相信它可能是真的——独裁政治会变得在道德上过时。它们会变成在道德上不可行的政府形式，而它创造的危机足以迫使我们找到另一条路。
**Dario Amodei:** That's the implication of this world. Let me just go on, because I think we will get diffusion eventually. The other concern I have is that governments will oppress their own people with AI. I'm worried about a world where you have a country in which there's already a government that's building a high-tech authoritarian state. To be clear, this is about the government. This is not about the people. We need to find a way for people everywhere to benefit. My worry here is about governments. My worry is if the world gets carved up into two pieces, one of those two pieces could be authoritarian or totalitarian in a way that's very difficult to displace.
Now, will governments eventually get powerful AI, and is there a risk of authoritarianism? Yes. Will governments eventually get powerful AI, and is there a risk of bad equilibria? Yes, I think both things. But the initial conditions matter. At some point, we're going to need to set up the rules of the road. I'm not saying that one country, either the United States or a coalition of democracies—which I think would be a better setup, although it requires more international cooperation than we currently seem to want to make—should just say, "These are the rules of the road." There's going to be some negotiation. The world is going to have to grapple with this. What I would like is for the democratic nations of the world—those whose governments represent closer to pro-human values—are holding the stronger hand and have more leverage when the rules of the road are set. So I'm very concerned about that initial condition.
Now, will governments eventually get powerful AI, and is there a risk of authoritarianism? Yes. Will governments eventually get powerful AI, and is there a risk of bad equilibria? Yes, I think both things. But the initial conditions matter. At some point, we're going to need to set up the rules of the road. I'm not saying that one country, either the United States or a coalition of democracies—which I think would be a better setup, although it requires more international cooperation than we currently seem to want to make—should just say, "These are the rules of the road." There's going to be some negotiation. The world is going to have to grapple with this. What I would like is for the democratic nations of the world—those whose governments represent closer to pro-human values—are holding the stronger hand and have more leverage when the rules of the road are set. So I'm very concerned about that initial condition.
**Dwarkesh Patel：** 我觉得这里确实有一个很难的问题，我不确定你怎么解决。我们在历史上不得不以某种方式站队。在 70 年代和 80 年代和中国的关系中，我们决定即使它是一个威权体制，我们也会和它接触。我觉得回顾起来那是正确的决定，因为它是一个国家威权体制，但十几亿人比否则要富裕得多、好得多。不清楚它否则是否会停止成为一个威权国家。你可以把朝鲜作为一个例子。我不认为保持一个威权国家以巩固自己权力需要那么多智慧。你可以想象一个朝鲜，拥有一个比所有人都差得多的 AI，但仍然足以维持权力。总的来说，似乎我们应该有这样一种态度：AI 的好处——以所有这些对人类的赋权和健康的形式——会很大。从历史上看，我们已经决定广泛传播技术的好处是好的，即使传播给政府是威权的人民。这是一个很难的问题，用 AI 来思考如何做，但历史上我们说过，"是的，这是一个正和世界，扩散技术仍然是值得的。"
**Dwarkesh Patel:** intelligence to remain an authoritarian country that continues to coalesce its own power. You can imagine a North Korea with an AI that's much worse than everybody else's, but still enough to keep power. In general, it seems like we should just have this attitude that the benefits of AI—in the form of all these empowerments of humanity and health—will be big. Historically, we have decided it's good to spread the benefits of technology widely, even to people whose governments are authoritarian. It is a tough question, how to think about it with AI, but historically we have said, "yes, this is a positive-sum world, and it's still worth diffusing the technology."
**Dario Amodei：** 我们有很多选择。把这框定为国家对国家的国家安全决策是一个视角，但还有很多其他视角。你可以想象一个世界，我们生产出了所有这些疾病的治愈方法。治愈方法可以卖给威权国家，但数据中心不行。芯片和数据中心不行，AI 行业本身也不行。另一个可能性是我觉得人们应该思考的。我们能否做出某些发展——要么作为 AI 的自然结果自然发生，要么我们可以通过在 AI 上构建技术来实现——创造一种均衡，使得威权国家无法拒绝其人民私人使用技术好处？有没有均衡是我们可以给威权国家的每个人自己的 AI 模型来保护他们免受监控，而威权国家没有办法在保持权力的同时打压这个？我不知道。
**Dario Amodei:** There are a number of choices we have. Framing this as a government-to-government decision in national security terms is one lens, but there are a lot of other lenses. You could imagine a world where we produce all these cures to diseases. The cures are fine to sell to authoritarian countries, but the data centers just aren't. The chips and the data centers aren't, and the AI industry itself isn't. Another possibility I think folks should think about is this. Could there be developments we can make—either that naturally happen as a result of AI, or that we could make happen by building technology on AI—that create an equilibrium where it becomes infeasible for authoritarian countries to deny their people private use of the benefits of the technology? Are there equilibria where we can give everyone in an authoritarian country their own AI model that defends them from surveillance and there isn't a way for the authoritarian country to crack down on this while retaining power? I don't know.
**Dwarkesh Patel：** 在我听来，如果走得够远的话，那将是一个威权国家会从内部瓦解的理由。
**Dwarkesh Patel:** That sounds to me like if that went far enough, it would be a reason why authoritarian countries would disintegrate from the inside.
**Dario Amodei：** 但也许有一个中间世界，有一种均衡，如果威权主义者想保持权力，他们就不能拒绝个性化的技术访问。但我确实对更激进的版本怀有希望。有没有可能技术本身就有——或者通过以某种方式在其上构建我们可以创造——对威权结构有这种溶解效应的属性？我们最初希望——回想 Obama 政府初期——社交媒体和互联网会有那种属性，结果发现没有。但如果我们能带着关于多少事情可能出错的知识再试一次，而且这是一种不同的技术呢？我不知道它是否会成功，但值得一试。只是非常不可预测。有第一性原理的理由说明威权主义可能被优待。一切都非常不可预测。我们只能认识到这个问题，想出 10 件可以尝试的事情，尝试它们，然后评估哪些在起作用，如果有的话。然后如果旧的不起作用就尝试新的。
**Dario Amodei:** But maybe there's a middle world where there's an equilibrium where, if they want to hold on to power, the authoritarians can't deny individualized access to the technology. But I actually do have a hope for the more radical version. Is it possible that the technology might inherently have properties—or that by building on it in certain ways we could create properties—that have this dissolving effect on authoritarian structures? Now, we hoped originally—think back to the beginning of the Obama administration—that social media and the internet would have that property, and it turns out not to. But what if we could try again with the knowledge of how many things could go wrong, and that this is a different technology? I don't know if it would work, but it's worth a try. It's just very unpredictable. There are first principles reasons why authoritarianism might be privileged. It's all very unpredictable. We just have to recognize the problem and come up with 10 things we can try, try those, and then assess which ones are working, if any. Then try new ones if the old ones aren't working.
**Dwarkesh Patel：** 但我猜这归结到今天，如你所说，我们不会向中国出售数据中心或芯片以及制造芯片的能力。所以在某种意义上，你确实在拒绝……中国经济、中国人民等等会从中获得一些好处，因为我们在这样做。然后美国经济也会有好处，因为这是一个正和世界。我们可以贸易。他们可以让他们国家的数据中心做一件事。我们可以让我们的做另一件事。你已经在说那种正和红利不值得去赋权那些国家？
**Dwarkesh Patel:** But I guess that nets out to today, as you say, that we will not sell data centers, or chips, and the ability to make chips to China. So in some sense, you are denying… There would be some benefits to the Chinese economy, Chinese people, et cetera, because we're doing that. Then there'd also be benefits to the American economy because it's a positive-sum world. We could trade. They could have their country's data centers doing one thing. We could have ours doing another. Already, you're saying it's not worth that positive-sum stipend to empower those countries?
**Dario Amodei：** 我要说的是，我们即将进入一个增长和经济价值会非常容易获得的世界，如果我们能够建造这些强大的 AI 模型的话。不会容易获得的是好处的分配、财富的分配、政治自由。这些是将来难以实现的事情。所以当我想到政策时，我认为技术和市场会提供所有基本的好处，这是我的根本信念，几乎快于我们能消化的速度。关于分配、政治自由和权利的这些问题才是真正重要的，政策应该聚焦于此。
**Dario Amodei:** What I would say is that we are about to be in a world where growth and economic value will come very easily if we're able to build these powerful AI models. What will not come easily is distribution of benefits, distribution of wealth, political freedom. These are the things that are going to be hard to achieve. So when I think about policy, I think that the technology and the market will deliver all the fundamental benefits, this is my fundamental belief, almost faster than we can take them. These questions about distribution and political freedom and rights are the ones that will actually matter and that policy should focus on.
**Dwarkesh Patel：** 说到分配，如你所提到的，我们有发展中国家。在很多情况下，追赶增长比我们希望的要弱。但当追赶增长确实发生时，根本原因是它们有未被充分利用的劳动力。我们可以把资本和专业知识从发达国家带到这些国家，然后它们可以相当快速地增长。显然，在一个劳动力不再是制约因素的世界里，这个机制不再起作用。所以希望基本上是依靠那些从 AI 中立即致富的人或国家的慈善事业吗？希望是什么？
**Dwarkesh Patel:** Speaking of distribution, as you were mentioning, we have developing countries. In many cases, catch-up growth has been weaker than we would have hoped for. But when catch-up growth does happen, it's fundamentally because they have underutilized labor. We can bring the capital and know-how from developed countries to these countries, and then they can grow quite rapidly. Obviously, in a world where labor is no longer the constraining factor, this mechanism no longer works. So is the hope basically to rely on philanthropy from the people or countries who immediately get wealthy from AI? What is the hope?
**Dario Amodei：** 慈善事业当然应该像过去一样发挥一些作用。但我认为如果我们能让增长内生化，增长总是更好更强的。在 AI 驱动的世界中，相关产业是什么？我说了我们不应该在中国建数据中心，但没有理由不在非洲建数据中心。事实上，我认为在非洲建数据中心会很好。只要它们不是中国拥有的，我们就应该在非洲建数据中心。我认为那是一件很好的事。没有理由不能建立一个 AI 驱动的制药产业。如果 AI 在加速药物发现，那么会有一堆生物技术创业公司。让我们确保其中一些发生在发展中世界。当然，在过渡期——我们可以谈谈人类没有角色的那个时点——人类仍然会在创办这些公司和监督 AI 模型方面有一些角色。所以让我们确保其中一些人在发展中世界，这样快速增长也能在那里发生。
**Dario Amodei:** Philanthropy should obviously play some role, as it has in the past. But I think growth is always better and stronger if we can make it endogenous. What are the relevant industries in an AI-driven world? I said we shouldn't build data centers in China, but there's no reason we shouldn't build data centers in Africa. In fact, I think it'd be great to build data centers in Africa. As long as they're not owned by China, we should build data centers in Africa. I think that's a great thing to do. There's no reason we can't build a pharmaceutical industry that's AI-driven. If AI is accelerating drug discovery, then there will be a bunch of biotech startups. Let's make sure some of those happen in the developing world. Certainly, during the transition—we can talk about the point where humans have no role—humans will still have some role in starting up these companies and supervising the AI models. So let's make sure some of those humans are in the developing world so that fast growth can happen there as well.
**Dwarkesh Patel：** 你们最近宣布 Claude 将有一部宪法，对齐到一组价值观，而不一定只是对齐到终端用户。我可以想象一个如果对齐到终端用户的世界，它会保持我们今天世界中存在的权力平衡，因为每个人都能拥有自己的 AI 为自己代言。坏行为者和好行为者的比例保持不变。这对我们今天的世界似乎是有效的。为什么不那样做，而是让 AI 携带一套特定的价值观更好？
**Dwarkesh Patel:** You guys recently announced that Claude is going to have a constitution that's aligned to a set of values, and not necessarily just to the end user. There's a world I can imagine where if it is aligned to the end user, it preserves the balance of power we have in the world today because everybody gets to have their own AI that's advocating for them. The ratio of bad actors to good actors stays constant. It seems to work out for our world today. Why is it better not to do that, but to have a specific set of values that the AI should carry forward?
**Dario Amodei：** 我不确定我会那样划分区别。这里可能有两个相关的区别。我认为你谈的是两者的混合。一个是，我们应该给模型一套"做这个"与"不做那个"的指令吗？另一个是，我们应该给模型一套行为原则吗？这纯粹是一个实践和经验的事情。通过教模型原则，让它从原则中学习，它的行为更一致，更容易覆盖边缘情况，模型更有可能做到人们想要它做的事。换句话说，如果你给它一个规则列表——"不要告诉人们怎么偷车、不要说韩语"——它并不真正理解这些规则，而且很难从中泛化。它只是一个做与不做的清单。而如果你给它原则——它有一些硬边界，比如"不要制造生物武器"，但——总体上你试图理解它应该追求什么，它应该如何运作。所以从实践角度来说，原则被证明是一种更有效的训练模型的方式。那是规则与原则的权衡。然后还有另一个你在谈论的东西，就是顺从性与内在动机的权衡。模型应该在多大程度上是一种"皮囊"，直接遵循给它指令的人的指示，还是模型应该有一套内在的价值观，自己去做事情？在那里我实际上会说，关于模型的一切都更偏向于它应该大部分做人们想要的事情。它应该大部分遵循指令。我们不是在试图构建一个自己跑去治理世界的东西。我们实际上相当偏向顺从的一方。现在，我们确实说有些事情模型不会做。我认为我们在宪法中以各种方式说过，在正常情况下，如果有人要求模型做一项任务，它应该做那项任务。那应该是默认。但如果你要求它做一些危险的事，或者伤害别人的事，那模型是不愿意做的。所以我实际上把它看作一个基本上顺从的模型，但有一些限制，而这些限制基于原则。
**Dario Amodei:** I'm not sure I'd quite draw the distinction in that way. There may be two relevant distinctions here. I think you're talking about a mix of the two. One is, should we give the model a set of instructions about "do this" versus "don't do this"? The other is, should we give the model a set of principles for how to act? It's kind of purely a practical and empirical thing that we've observed. By teaching the model principles, getting it to learn from principles, its behavior is more consistent, it's easier to cover edge cases, and the model is more likely to do what people want it to do. In other words, if you give it a list of rules—"don't tell people how to hot-wire a car, don't speak in Korean"—it doesn't really understand the rules, and it's hard to generalize from them. It's just a list of do's and don't's. Whereas if you give it principles—it has some hard guardrails like "Don't make biological weapons" but—overall you're trying to understand what it should be aiming to do, how it should be aiming to operate. So just from a practical perspective, that turns out to be a more effective way to train the model. That's the rules versus principles trade-off. Then there's another thing you're talking about, which is the corrigibility versus intrinsic motivation trade-off. How much should the model be a kind of "skin suit" where it just directly follows the instructions given to it by whoever is giving those instructions, versus how much should the model have an inherent set of values and go off and do things on its own? There I would actually say everything about the model is closer to the direction that it should mostly do what people want. It should mostly follow instructions. We're not trying to build something that goes off and runs the world on its own. We're actually pretty far on the corrigible side. Now, what we do say is there are certain things that the model won't do. I think we say it in various ways in the constitution, that under normal circumstances, if someone asks the model to do a task, it should do that task. That should be the default. But if you've asked it to do something dangerous, or to harm someone else, then the model is unwilling to do that. So I actually think of it as a mostly corrigible model that has some limits, but those limits are based on principles.
**Dwarkesh Patel：** 那么根本性的问题是，这些原则是如何确定的？这不是 Anthropic 特有的问题。对任何 AI 公司都是一样的。但因为你们是实际上写下了原则的人，我有机会问你这个问题。通常，宪法是写下来的，固定的，有一个更新和修改的过程等等。在这种情况下，它似乎是一份 Anthropic 的人写的文件，可以随时更改，它指导着将成为大量经济活动基础的系统的行为。你怎么想这些原则应该如何被确定？
**Dwarkesh Patel:** Then the fundamental question is, how are those principles determined? This is not a special question for Anthropic. This would be a question for any AI company. But because you have been the ones to actually write down the principles, I get to ask you this question. Normally, a constitution is written down, set in stone, and there's a process of updating it and changing it and so forth. In this case, it seems like a document that people at Anthropic write, that can be changed at any time, that guides the behavior of systems that are going to be the basis of a lot of economic activity. How do you think about how those principles should be set?
**Dario Amodei：** 我认为这里可能有三种大小的循环，三种迭代方式。一是我们在 Anthropic 内部迭代。我们训练模型，我们不满意，我们修改宪法。我认为这样做是好的。偶尔发布宪法的公开更新是好的，因为人们可以评论。第二层循环是不同公司有不同的宪法。我认为这很有用。Anthropic 发布一部宪法，Gemini 发布一部宪法，其他公司发布一部宪法。人们可以看它们并比较。外部观察者可以批评说，"我喜欢这部宪法的这个东西，那部宪法的那个东西。"这为所有公司创造了一个软激励和反馈，去取各家之长并改进。然后我认为有第三个循环，就是 AI 公司以外的社会，以及不仅仅是那些没有硬权力的评论者。在那里我们做了一些实验。几年前，我们和 Collective Intelligence Project 做了一个实验，基本上是调查人们，问他们我们的 AI 宪法里应该有什么。当时，我们纳入了其中一些改变。所以你可以想象用我们对宪法采取的新方法来做类似的事情。这更难一些，因为当宪法是一个做与不做的列表时，那种方法更容易采用。在原则层面，它必须有一定的一致性。但你仍然可以想象从各种各样的人那里获取意见。你也可以想象——这是一个疯狂的想法，但这整个采访都是关于疯狂想法的——代议制政府的体系有输入。我今天不会这么做，因为立法过程太慢了。这恰恰是为什么我认为我们应该对立法过程和 AI 监管持谨慎态度。但原则上没有理由你不能说，"所有 AI 模型必须有一部以这些东西开头的宪法，然后你可以在后面附加其他东西，但必须有这个优先的特殊部分。"我不会那样做。那太僵硬了，听起来过分规定性，我认为过分激进的立法就是那样。但那是你可以尝试做的一件事。有没有某种不那么重手的版本？也许有。
**Dario Amodei:** I think there are maybe three sizes of loop here, three ways to iterate. One is we iterate within Anthropic. We train the model, we're not happy with it, and we change the constitution. I think that's good to do. Putting out public updates to the constitution every once in a while is good because people can comment on it. The second level of loop is different companies having different constitutions. I think it's useful. Anthropic puts out a constitution, Gemini puts out a constitution, and other companies put out a constitution. People can look at them and compare. Outside observers can critique and say, "I like this thing from this constitution and this thing from that constitution." That creates a soft incentive and feedback for all the companies to take the best of each element and improve. Then I think there's a third loop, which is society beyond the AI companies and beyond just those who comment without hard power. There we've done some experiments. A couple years ago, we did an experiment with the Collective Intelligence Project to basically poll people and ask them what should be in our AI constitution. At the time, we incorporated some of those changes. So you could imagine doing something like that with the new approach we've taken to the constitution. It's a little harder because it was an easier approach to take when the constitution was a list of dos and don'ts. At the level of principles, it has to have a certain amount of coherence. But you could still imagine getting views from a wide variety of people. You could also imagine—and this is a crazy idea, but this whole interview is about crazy ideas—systems of representative government having input. I wouldn't do this today because the legislative process is so slow. This is exactly why I think we should be careful about the legislative process and AI regulation. But there's no reason you couldn't, in principle, say, "All AI models have to have a constitution that starts with these things, and then you can append other things after it, but there has to be this special section that takes precedence." I wouldn't do that. That's too rigid and sounds overly prescriptive in a way that I think overly aggressive legislation is. But that is a thing you could try to do. Is there some much less heavy-handed version of that? Maybe.
**Dwarkesh Patel：** 我真的很喜欢第二个控制循环。显然，这不是实际政府的宪法应该或确实运作的方式。不存在那种最高法院会感受人们的感觉——什么是氛围——然后据此更新宪法的模糊感觉。对于实际政府，有一个更正式的程序过程。但你有一个宪法之间竞争的愿景，这实际上非常让人想起一些自由主义者的特许城市人士过去谈论的那种不同政府的群岛会是什么样子。它们之间会有选择压力，看谁能最有效地运作，人们在哪里最幸福。在某种意义上，你在重新创造那种群岛式乌托邦的愿景。
**Dwarkesh Patel:** I really like control loop two. Obviously, this is not how constitutions of actual governments do or should work. There's not this vague sense in which the Supreme Court will feel out how people are feeling—what are the vibes—and update the constitution accordingly. With actual governments, there's a more formal, procedural process. But you have a vision of competition between constitutions, which is actually very reminiscent of how some libertarian charter cities people used to talk, about what an archipelago of different kinds of governments would look like. There would be selection among them of who could operate the most effectively and where people would be the happiest. In a sense, you're recreating that vision of a utopia of archipelagos.
**Dario Amodei：** 我认为那个愿景有值得推荐的地方，也有会出错的地方。这是一个有趣的、在某些方面引人注目的愿景，但会有你没想象到的事情出错。所以我也喜欢循环二，但我觉得整个事情必须是循环一、二和三的某种混合，这是一个比例的问题。我认为那必须是答案。
**Dario Amodei:** I think that vision has things to recommend it and things that will go wrong with it. It's an interesting, in some ways compelling, vision, but things will go wrong that you hadn't imagined. So I like loop two as well, but I feel like the whole thing has got to be some mix of loops one, two, and three, and it's a matter of the proportions. I think that's gotta be the answer.
**Dwarkesh Patel：** 当有人最终写出这个时代的《原子弹的制造》时，从历史记录中最难发掘出来的、他们最有可能错过的是什么？
**Dwarkesh Patel:** When somebody eventually writes the equivalent of The Making of the Atomic Bomb for this era, what is the thing that will be hardest to glean from the historical record that they're most likely to miss?
**Dario Amodei：** 我认为有几件事。一是，在这条指数曲线的每一刻，外界对它的不理解程度。这是历史中经常存在的偏见。任何实际发生的事情在回顾时看起来都是不可避免的。当人们回顾时，他们很难把自己放在那些实际上在对这件事是否会发生下注的人的位置上——那件事并非不可避免，我们有这些论证，比如我为 scaling 做出的论证，或者持续学习会被解决。我们内部有些人对这件事发生的概率很高，但我们外面有一个完全没有在此基础上行动的世界。我认为这件事的怪异之处，不幸的是它的封闭性……如果我们距离它发生只有一两年，街上的普通人对此毫无概念。这是我试图通过备忘录、通过和政策制定者交谈来改变的事情之一。我不知道，但我觉得这只是一件疯狂的事。最后，我会说——这可能适用于几乎所有历史危机时刻——一切发生得有多快，一切同时发生。你可能以为那些经过深思熟虑的决定，实际上你必须做那个决定，然后你同一天还得做 30 个其他决定，因为一切都发生得太快了。你甚至不知道哪些决定最终会是有后果的。我的一个担忧——虽然它也是对正在发生的事情的洞察——是某个非常关键的决定会是某个人走进我的办公室说，"Dario，你有两分钟。我们应该做 A 还是 B？"某人给我一份随机的半页备忘录问，"我们做 A 还是 B？"我说，"我不知道。我得去吃午饭。做 B 吧。"那最终成为了有史以来最有后果的事情。
**Dario Amodei:** I think a few things. One is, at every moment of this exponential, the extent to which the world outside it didn't understand it. This is a bias that's often present in history. Anything that actually happened looks inevitable in retrospect. When people look back, it will be hard for them to put themselves in the place of people who were actually making a bet on this thing to happen that wasn't inevitable, that we had these arguments like the arguments I make for scaling or that continual learning will be solved. Some of us internally put a high probability on this happening, but there's a world outside us that's not acting on that at all. I think the weirdness of it, unfortunately the insularity of it... If we're one year or two years away from it happening, the average person on the street has no idea. That's one of the things I'm trying to change with the memos, with talking to policymakers. I don't know but I think that's just a crazy thing. Finally, I would say—and this probably applies to almost all historical moments of crisis—how absolutely fast it was happening, how everything was happening all at once. Decisions that you might think were carefully calculated, well actually you have to make that decision, and then you have to make 30 other decisions on the same day because it's all happening so fast. You don't even know which decisions are going to turn out to be consequential. One of my worries—although it's also an insight into what's happening—is that some very critical decision will be some decision where someone just comes into my office and is like, "Dario, you have two minutes. Should we do thing A or thing B on this?" Someone gives me this random half-page memo and asks, "Should we do A or B?" I'm like, "I don't know. I have to eat lunch. Let's do B." That ends up being the most consequential thing ever.
**Dwarkesh Patel：** 最后一个问题。没有科技 CEO 通常每隔几个月写 50 页的备忘录。你似乎为自己和围绕你的公司建立了一种与这种更偏知识分子型 CEO 角色兼容的方式。我想了解你是怎么构建的。这是怎么运作的？你是不是消失几周然后告诉公司"这是备忘录。这是我们要做的"？据报道你在内部也写了很多这些。
**Dwarkesh Patel:** So final question. There aren't tech CEOs who are usually writing 50-page memos every few months. It seems like you have managed to build a role for yourself and a company around you which is compatible with this more intellectual-type role of CEO. I want to understand how you construct that. How does that work? Do you just go away for a couple of weeks and then you tell your company, "This is the memo. Here's what we're doing"? It's also reported that you write a bunch of these internally.
**Dario Amodei：** 对于这篇特定的文章，我是在寒假期间写的。我很难找到时间来写它。但我以更广泛的方式思考这个问题。我认为这关系到公司的文化。我大概花三分之一，也许 40% 的时间确保 Anthropic 的文化是好的。随着 Anthropic 变大，直接参与模型的训练、模型的发布、产品的构建变得更难了。公司有 2500 人。我有某些直觉，但要参与每一个细节非常困难。我尽可能多地参与，但有一件事非常有杠杆性，就是确保 Anthropic 是一个好的工作场所，人们喜欢在那里工作，每个人都把自己当作团队成员，每个人一起工作而不是互相对抗。我们看到随着一些其他 AI 公司的成长——不指名道姓——我们开始看到失去凝聚力和内斗。我会说甚至从一开始就有很多那样的情况，但它变得更严重了。我认为我们在保持公司团结方面做得非常好，即使不完美——让每个人感受到使命，我们对使命是真诚的，每个人都相信其他人在那里是为了正确的原因。我们是一个团队，人们没有在试图损人利己或互相捅刀，我认为这在其他一些地方经常发生。
**Dario Amodei:** For this particular one, I wrote it over winter break. I was having a hard time finding the time to actually write it. But I think about this in a broader way. I think it relates to the culture of the company. I probably spend a third, maybe 40%, of my time making sure the culture of Anthropic is good. As Anthropic has gotten larger, it's gotten harder to get directly involved in the training of the models, the launch of the models, the building of the products. It's 2,500 people. I have certain instincts, but it's very difficult to get involved in every single detail. I try as much as possible, but one thing that's very leveraged is making sure Anthropic is a good place to work, people like working there, everyone thinks of themselves as team members, and everyone works together instead of against each other. We've seen as some of the other AI companies have grown—without naming any names—we're starting to see decoherence and people fighting each other. I would argue there was even a lot of that from the beginning, but it's gotten worse. I think we've done an extraordinarily good job, even if not perfect, of holding the company together, making everyone feel the mission, that we're sincere about the mission, and that everyone has faith that everyone else there is working for the right reason. That we're a team, that people aren't trying to get ahead at each other's expense or backstab each other, which again, I think happens a lot at some of the other places.
**Dwarkesh Patel：** 你是怎么做到这一点的？
**Dwarkesh Patel:** How do you make that the case?
**Dario Amodei：** 很多因素。是我，是 Daniela——她负责公司的日常运营，是联合创始人们，是我们雇佣的其他人，是我们试图创造的环境。但我认为文化中重要的一点是，其他领导者也一样，但尤其是我，必须清楚阐述公司是关于什么的，为什么在做它正在做的事，它的战略是什么，它的价值观是什么，它的使命是什么，它代表什么。当你到了 2500 人的时候，你不能一个一个来做这件事。你必须写，或者你必须对全公司讲话。这就是为什么我每两周站在全公司面前讲一个小时。我不会说我在内部写文章。我做两件事。一，我写一个叫 DVQ 的东西，Dario Vision Quest。不是我给它起的这个名字。那是它得到的名字，这是一个我试图反对的名字，因为它让人听起来像我跑去抽迷幻药了或什么的。但这个名字就是留下来了。所以我每两周站在全公司面前。我有一份三四页的文件，我就讲三四个关于内部正在发生什么的不同话题，我们正在生产的模型，产品，外部行业，整个世界与 AI 和地缘政治相关的情况。就是这些的混合。我非常诚实地说，"这是我在想的，这是 Anthropic 领导层在想的"，然后我回答问题。这种直接联系有很多价值，当你要通过六层深的层级传递信息时很难实现。公司很大一部分人会来参加，无论是现场还是线上。这真的意味着你可以传达很多东西。
我做的另一件事是我在 Slack 上有一个频道，我在那里写很多东西并且经常评论。通常那是对我在公司看到的事情或人们问的问题的回应。我们做内部调查，有人们担心的事情，所以我会写出来。我对这些事情非常诚实。我就很直接地说。关键是要建立起这样的声誉：告诉公司正在发生什么的真相，直呼其名，承认问题，避免那种企业腔调——那种在公众场合经常必要的防御性沟通，因为世界很大，充满了恶意解读的人。但如果你有一家你信任的人组成的公司，我们试图雇佣我们信任的人，那你真的可以完全不加过滤。我认为那是公司的巨大优势。它使公司成为一个更好的工作场所，让人们超越各部分之和，增加我们完成使命的可能性，因为每个人都对使命有相同的理解，每个人都在辩论和讨论如何最好地完成使命。
我做的另一件事是我在 Slack 上有一个频道，我在那里写很多东西并且经常评论。通常那是对我在公司看到的事情或人们问的问题的回应。我们做内部调查，有人们担心的事情，所以我会写出来。我对这些事情非常诚实。我就很直接地说。关键是要建立起这样的声誉：告诉公司正在发生什么的真相，直呼其名，承认问题，避免那种企业腔调——那种在公众场合经常必要的防御性沟通，因为世界很大，充满了恶意解读的人。但如果你有一家你信任的人组成的公司，我们试图雇佣我们信任的人，那你真的可以完全不加过滤。我认为那是公司的巨大优势。它使公司成为一个更好的工作场所，让人们超越各部分之和，增加我们完成使命的可能性，因为每个人都对使命有相同的理解，每个人都在辩论和讨论如何最好地完成使命。
**Dario Amodei:** It's a lot of things. It's me, it's Daniela, who runs the company day to day, it's the co-founders, it's the other people we hire, it's the environment we try to create. But I think an important thing in the culture is that the other leaders as well, but especially me, have to articulate what the company is about, why it's doing what it's doing, what its strategy is, what its values are, what its mission is, and what it stands for. When you get to 2,500 people, you can't do that person by person. You have to write, or you have to speak to the whole company. This is why I get up in front of the whole company every two weeks and speak for an hour. I wouldn't say I write essays internally. I do two things. One, I write this thing called a DVQ, Dario Vision Quest. I wasn't the one who named it that. That's the name it received, and it's one of these names that I tried to fight because it made it sound like I was going off and smoking peyote or something. But the name just stuck. So I get up in front of the company every two weeks. I have a three or four-page document, and I just talk through three or four different topics about what's going on internally, the models we're producing, the products, the outside industry, the world as a whole as it relates to AI and geopolitically in general. Just some mix of that. I go through very honestly and I say, "This is what I'm thinking, and this is what Anthropic leadership is thinking," and then I answer questions. That direct connection has a lot of value that is hard to achieve when you're passing things down the chain six levels deep. A large fraction of the company comes to attend, either in person or virtually. It really means that you can communicate a lot. The other thing I do is I have a channel in Slack where I just write a bunch of things and comment a lot. Often that's in response to things I'm seeing at the company or questions people ask. We do internal surveys and there are things people are concerned about, and so I'll write them up. I'm just very honest about these things. I just say them very directly. The point is to get a reputation of telling the company the truth about what's happening, to call things what they are, to acknowledge problems, to avoid the sort of corpo speak, the kind of defensive communication that often is necessary in public because the world is very large and full of people who are interpreting things in bad faith. But if you have a company of people who you trust, and we try to hire people that we trust, then you can really just be entirely unfiltered. I think that's an enormous strength of the company. It makes it a better place to work, it makes people more than the sum of their parts, and increases the likelihood that we accomplish the mission because everyone is on the same page about the mission, and everyone is debating and discussing how best to accomplish the mission.
**Dwarkesh Patel：** 好吧，既然没有对外的 Dario Vision Quest，我们就有这次采访。这次采访有点像那样。这很有趣，Dario。感谢你做这次访谈。
**Dwarkesh Patel:** Well, in lieu of an external Dario Vision Quest, we have this interview. This interview is a little like that. This has been fun, Dario. Thanks for doing it.
**Dario Amodei：** 谢谢你，Dwarkesh。
**Dario Amodei:** Thank you, Dwarkesh.
