Sam Altman:Dots如何夺回时间,AI应走文艺复兴而非工业革命之路Sam Altman on How Dots Reclaim Time and Why AI Must Be a Renaissance, Not an Industrial Revolution
The Takeaway:AI不应把人变成机器上的齿轮,而应让个人重新掌控时间与创造力,Dots就是最好的证明。
OpenAI CEO Sam Altman在Every播客中分享了他如何用新推出的Dots(个人AI代理)重新夺回清晨深度工作时间。以前他一醒来就被迫检查手机,大脑被过夜紧急事件污染;现在Dots会用判断力过滤信息,只在真正紧急时打断,否则自行处理,让他能陪孩子或专注创作。Dots甚至能在Dev Day keynote前五分钟提醒他demo可能出问题,或在他放弃搜索后通宵找到Slack截图。
Altman强调,人们其实不关心机器做了什么,只关心人。一万个agent协作解决Navier-Stokes方程,公众讨论的却是参与的人。他对比文艺复兴与工业革命:前者以人为中心,后者把人变成专业齿轮。OpenAI要走前者——用平台降低创业门槛,让一人公司涌现,而不是成为唯一巨头。Dev Day一口气发布22项更新(包括Space协作文档、Decisions API、MCP服务器、订阅可携等),就是在实践“我们自己也必须用这些工具做出更多东西”。
“七个人的Dots一起工作,能创造出一种我们头脑中都存在、却从未成形的涌现事物。”这正是他眼中人机协作的未来。
OpenAI CEO Sam Altman在Every播客中分享了他如何用新推出的Dots(个人AI代理)重新夺回清晨深度工作时间。以前他一醒来就被迫检查手机,大脑被过夜紧急事件污染;现在Dots会用判断力过滤信息,只在真正紧急时打断,否则自行处理,让他能陪孩子或专注创作。Dots甚至能在Dev Day keynote前五分钟提醒他demo可能出问题,或在他放弃搜索后通宵找到Slack截图。
Altman强调,人们其实不关心机器做了什么,只关心人。一万个agent协作解决Navier-Stokes方程,公众讨论的却是参与的人。他对比文艺复兴与工业革命:前者以人为中心,后者把人变成专业齿轮。OpenAI要走前者——用平台降低创业门槛,让一人公司涌现,而不是成为唯一巨头。Dev Day一口气发布22项更新(包括Space协作文档、Decisions API、MCP服务器、订阅可携等),就是在实践“我们自己也必须用这些工具做出更多东西”。
“七个人的Dots一起工作,能创造出一种我们头脑中都存在、却从未成形的涌现事物。”这正是他眼中人机协作的未来。
The Takeaway: AI should not turn people into cogs in a machine; it should hand control of time and creativity back to individuals—and Dots is the clearest proof yet.
OpenAI CEO Sam Altman explained how the newly launched Dots (personal AI agents) restored his early-morning deep-work window. He used to check his phone the moment he woke up and pollute his mind with overnight fires; now Dots exercises judgment, surfaces only true emergencies, and handles the rest so he can stay with his kids or do creative work. Dots even pinged him five minutes before a live Dev Day demo warning that a production issue would break it, and later found a Slack screenshot he had given up searching for after spending the night looking.
Altman keeps returning to a simple observation: humans do not care what machines do; we care about the people behind the art, the company, or the idea. When 10,000 agents solved Navier-Stokes in perfect coordination, almost none of the public reaction was about the math—it was about the humans involved. He contrasts the Renaissance (people-centered) with the Industrial Revolution (machine-centered, turning people into specialized cogs). OpenAI is betting on the former: open the platform, lower the bar for one-person companies, and refuse to become the single megacorp. The 22 launches at Dev Day—including Space live documents, the Decisions API, MCP server deployment inside ChatGPT, and portable subscriptions—are the company practicing what it preaches: if the tools let people build more, OpenAI itself must ship more.
“Seven people’s dots working together can create something new that kind of existed in all of our heads to some degree, but was like an emergent thing.” That is the collaboration future he is racing toward.
查看原文 →
OpenAI CEO Sam Altman explained how the newly launched Dots (personal AI agents) restored his early-morning deep-work window. He used to check his phone the moment he woke up and pollute his mind with overnight fires; now Dots exercises judgment, surfaces only true emergencies, and handles the rest so he can stay with his kids or do creative work. Dots even pinged him five minutes before a live Dev Day demo warning that a production issue would break it, and later found a Slack screenshot he had given up searching for after spending the night looking.
Altman keeps returning to a simple observation: humans do not care what machines do; we care about the people behind the art, the company, or the idea. When 10,000 agents solved Navier-Stokes in perfect coordination, almost none of the public reaction was about the math—it was about the humans involved. He contrasts the Renaissance (people-centered) with the Industrial Revolution (machine-centered, turning people into specialized cogs). OpenAI is betting on the former: open the platform, lower the bar for one-person companies, and refuse to become the single megacorp. The 22 launches at Dev Day—including Space live documents, the Decisions API, MCP server deployment inside ChatGPT, and portable subscriptions—are the company practicing what it preaches: if the tools let people build more, OpenAI itself must ship more.
“Seven people’s dots working together can create something new that kind of existed in all of our heads to some degree, but was like an emergent thing.” That is the collaboration future he is racing toward.