Editorial illustration for Sam Altman Warns AI Could Replace Human Decision-Making
Altman Warns Against AI Replacing Human Decision-Making
Sam Altman spent part of Wednesday afternoon at The New York Times DealBook Summit pushing back on an idea he says has taken hold in parts of Silicon Valley: that artificial intelligence should eventually take over the big decisions currently made by people. The OpenAI CEO, appearing alongside co-founder and president Greg Brockman, framed the debate around what's sometimes called recursive self-improvement, the theory that AI systems will keep making smarter versions of themselves until they outpace human oversight entirely. Proponents of this "gradual handover" view argue that outcome is either inevitable or desirable, with AI eventually running major industries and institutions on its own.
Altman and Brockman don't buy it. Both men told the DealBook audience they're aligned on the core principle that humans need to remain in control as AI capabilities grow, even as they diverge on the specifics of how to get there and what obstacles will show up along the way. The exchange offered a rare look at how two of the people building the most advanced AI systems on the planet are thinking about the limits they want to put on their own creation.
OpenAI co-founder and CEO Sam Altman believes that people should stay in charge of the future when it comes to how AI technology impacts the world and everyone’s lives.
Why this matters
Altman and Brockman are naming a real fork in the road, but neither one offers much beyond the warning itself. If monitoring tools for frontier models still don't exist, as Brockman admits, then talk of "keeping humans in charge" is mostly aspirational. For developers and founders building on top of these systems, the gap between OpenAI's stated principles and its actual safety infrastructure is the thing to watch, not the DealBook soundbites.
Alignment research funding, red-teaming practices, and whatever new monitoring tech Brockman alludes to will tell us more about OpenAI's real priorities than any stage appearance. We'd also note the framing problem: "gradual handover to AI" treats this as a slow, manageable slope, when the actual risk is institutions adopting AI decision-making faster than the safety tooling matures. Researchers should press OpenAI publicly for specifics on what "new technology" for monitoring actually looks like, and on what timeline.
Until then, this is a company stating good intentions while admitting it hasn't built the infrastructure to back them up.
Common Questions Answered
What is Sam Altman's position on AI taking over human decision-making?
Sam Altman believes that people should remain in charge of major decisions and that AI should not eventually take over decision-making currently made by humans. He pushed back against the idea that has taken hold in parts of Silicon Valley that artificial intelligence should eventually assume control of big decisions. Altman framed this debate around the concept of recursive self-improvement, where AI systems theoretically make smarter versions of themselves.
What concern does Greg Brockman raise about monitoring AI systems?
Greg Brockman admits that monitoring tools for frontier models still don't exist, which undermines claims about keeping humans in charge of AI technology. This gap between stated principles and actual safety infrastructure represents a significant challenge in ensuring human oversight of AI development. Without proper monitoring tools in place, the ability to maintain human control over AI decision-making becomes largely aspirational rather than practical.
Why does the article suggest developers should focus on OpenAI's safety infrastructure rather than public statements?
The article argues that while Altman and Brockman identify a real fork in the road regarding AI autonomy, they offer little beyond warnings without concrete solutions. The gap between OpenAI's stated principles about keeping humans in charge and its actual safety infrastructure is what developers and founders building on these systems should closely monitor. Public soundbites at events like DealBook are less important than examining alignment research funding, red-teaming efforts, and real safety measures.
What is recursive self-improvement in the context of this article?
Recursive self-improvement is a theory that AI systems will continue making smarter versions of themselves until they surpass human intelligence and decision-making capabilities. Sam Altman used this concept to frame the debate about whether AI should eventually take over decisions currently made by people. This theory represents one of the key concerns driving the discussion about maintaining human control over AI technology.
Further Reading
- Sam Altman warns against a future where AI, not humans, calls the shots - TechCrunch
- Sam Altman’s remarks at the United Nations Security Council - OpenAI
- ChatGPT and Anthropic chiefs tell UN that humans must be at centre of AI decisions - The National
- Why these startup CEOs don't think AI will replace human roles - TechCrunch
- People should always be in control: Sam Altman warns AI progress must have limits - Moneycontrol