One of the most underestimated threats of artificial intelligence (AI) may not be the loss of jobs, the spread of deepfakes, or the spread of AI out of control, but the gradual weakening of human thinking. This warning was made by the head of OpenAI, Sam Altman, who admitted that excessive dependence on AI could lead to “cognitive atrophy” when people stop solving intellectual problems on their own.
On the Invest Like the Best podcast, Altman was asked what unsolved problems in artificial intelligence he considers the most important. Instead of the usual topics – the security of superintelligent systems, regulation or the impact of AI on the labor market – the head of OpenAI, which developed the popular chatbot ChatGPT, named a risk that, in his opinion, is almost not discussed.
“One of the things that I don’t think gets enough attention is how do we avoid cognitive atrophy?” Altman said.
The big question, he says, is whether people can use AI in a way that allows them to continue to develop their own thinking, rather than gradually shifting more and more intellectual work to algorithms.
“How do we use these tools in a way that allows us to continue to train our brains to understand the things that really matter?” – he noted.
Although such concerns are rarely heard from the heads of the largest AI companies, some researchers have been warning about this problem for several years. As Business Insider notes, Massachusetts Institute of Technology (MIT) researcher Natalia Kosmyna previously stated that generative AI is the first tool that can not just automate routine actions, but actually take over the thinking process itself. Preliminary results from her study showed that participants who regularly used AI to write essays performed worse over time than people who used regular Google searches or completed tasks on their own.
Theoretical neuroscientist and AI researcher Vivien Ming calls this phenomenon “chronic substitution.” According to her, if a person constantly delegates intellectual work to algorithms, this can gradually reduce the so-called cognitive reserve – one of the factors that helps the brain resist age-related changes and dementia.
However, Altman is not the only AI company executive to talk about such risks. Last year, the CEO of France's Mistral AI, Arthur Mensch, warned that one of the main threats could be deskilling if people began to rely too heavily on artificial intelligence to do their jobs.
At the same time, Altman himself does not consider such a scenario inevitable. In the same interview, he described a future in which personal AI would read documents, analyze meetings, remind important emails, and even suggest new ideas. However, in his opinion, there are qualities that modern models are not yet able to reproduce.
“The world may even need a new word for the type of judgment that humans are actually very good at but AI struggles with,” the OpenAI chief said.
