Sam Altman’s underrated AI fear: that we stop thinking for ourselves

OpenAI CEO Sam Altman warned on the podcast 'Invest Like the Best' that the biggest unaddressed risk of artificial intelligence is 'cognitive atrophy,' where people stop thinking for themselves as they rely on AI tools. Altman said the danger of laziness rather than apocalypse is one that 'doesn't get much attention,' even as his company pushes ChatGPT into more of daily life. Preliminary research by MIT's Nataliya Kosmyna found that people who used generative AI to write essays performed worse over time than those who used Google or no help, and theoretical neuroscientist Vivienne Ming warned that 'chronic substitution' of cognitive work could erode the brain's defense against dementia.

The man whose company is trying to get the world to use AI for everything has named a risk he thinks almost no one is talking about: that people will let it do so much of their thinking that their own minds go slack. Sam Altman, OpenAI’s chief executive, calls it “cognitive atrophy,” and puts it near the top of AI’s unresolved questions. He said it on the podcast “Invest Like the Best.” Asked what most worried him about AI’s future, Altman replied: “One that I don’t think gets much attention is how are we going to avoid cognitive atrophy?” His framing was less doom than discipline. “How are we going to use these tools and make sure that we are stretching our brains more and more and continuing to understand the stuff that really matters?” he asked, casting the danger as one of laziness rather than apocalypse. It is a striking worry coming from him. OpenAI’s whole business is persuading people to lean on ChatGPT https://thenextweb.com/news/when-the-machine-asks-you-to-stay for more of their work and their decisions, which makes its founder fretting about over-reliance a little like a confectioner warning about sugar. The research is starting to back the worry. Nataliya Kosmyna, an MIT researcher, ran a preliminary study in which people who used generative AI to write essays performed worse over time than those who used Google or no help at all. Her framing was blunt. “There’s never been a tool like this that has offered us the opportunity to let thinking be done for us.” Others put the risk in starker terms. Vivienne Ming, a theoretical neuroscientist, calls the habit of handing cognitive work to AI “chronic substitution,” and warns that repeating it could erode the cognitive reserve that helps defend the brain against dementia. The concern sharpens as AI takes on more. The tools have moved beyond answering questions to doing document review, research, analysis, customer support, and coding, which means the mental work being offloaded is no longer trivial. Altman is not the first tech boss to say so. Last year Arthur Mensch, the chief executive of the French lab Mistral, told The Times that AI’s biggest risk might be “deskilling” workers and making them progressively lazier as they lean on it to think. The pattern is old, only sharper. Calculators, satnav, and search each took over a skill people used to keep in their heads, and each drew the same warning, usually without the collapse the warnings predicted. What is different now is breadth. Earlier tools offloaded one narrow task; a general-purpose assistant can read, write, reason, and decide, which makes the question of what humans keep practising a far larger one. Altman, for his part, did not stay gloomy. He described a “pretty magical” future in which an always-on personal AI reads your documents, listens to your meetings, recalls emails from weeks ago, and uses spare overnight computing power to generate ideas and tell you what to do next. Even so, he drew a line around what the machines cannot do. “The world may need like a new kind of word for the kind of judgment that people are very good at, that AI seems to really deeply struggle with,” Altman said, a rare admission of the tool’s limits from its chief salesman. That is the tension he is living. Altman can believe both that AI will make people more capable and that it might make them lazier, and OpenAI is pushing the tool https://thenextweb.com/news/openai-agents-10-million-chatgpt-work into ever more of daily life while its founder warns about exactly that. The honest answer is that no one knows yet. Whether AI stretches our brains or lets them soften depends on how people use it, and Altman, who has worried before https://thenextweb.com/news/sam-altman-tiktok-addiction-deleted-openai-sora about his own inability to put down a screen, is not promising the good outcome, only naming the risk. Get the TNW newsletter Get the most important tech news in your inbox each week.