Altman Says OpenAI’s Unpublished Models Lead Public Models in Math
During a 45-minute interview with Matthew Berman, Sam Altman said OpenAI’s unpublished internal models had reached performance levels on some mathematics research tasks that widely available public models could not approach, citing the company’s recent mathematics work as an example. He said he hoped AI could help identify a meaningful candidate treatment for a serious disease within roughly two years, while noting that any treatment would still require testing and regulatory review. Altman also said software demand could leave programmers busier and potentially better paid, and that personal AI assistants would require more mature products, greater computing capacity and new hardware for continuous environmental awareness.
Altman’s remarks connect OpenAI’s internal model capability with broader claims about AI-assisted science, programming demand and the infrastructure needed for personal AI assistants. His treatment forecast and labor-market view remain his stated expectations.
如果有一天,AI 能够完成最复杂的数学证明、发现新的物理学规律,甚至找到重大疾病的治疗方法,人类还需要做什么?
在 OpenAI CEO Sam Altman 看来,这已经不再是一个遥远的问题。随着 AI 开始参与真正的科学发现,人类正在进入一个新阶段。他预计,未来几年内,AI 有望帮助找到重大疾病的候选治疗方案。但相比医学进展,最让他个人感到兴奋的,反而是物理学领域可能出现的重大突破。
不过,AI 能力越强,人类就一定会过得越好吗?
最近,在与 AI 博主 Matthew Berman 长达 45 分钟的对话中,奥特曼谈到了一个颇为矛盾的现实:AI 已经能够承担越来越多的工作,但人们似乎并没有因此变得清闲。
程序员就是一个典型例子。过去,人们普遍担心 AI 编程会让开发者失业,但奥特曼发现,程序员反而可能比以往更加忙碌,创造更大的价值,甚至赚到更多的钱。原因之一在于,人类对软件的需求,远远超出了此前的想象。