Altman warns: excessive AI can atrophy our thinking

What happened?

Sam Altman, CEO of OpenAI, has warned about a risk he considers undervalued: that people delegate so much of their thinking to artificial intelligence that their own cognitive abilities deteriorate. Altman calls this phenomenon “cognitive atrophy”. The statement came during an interview on Adam Grant's podcast ReThinking, published on March 24, 2025, and was reported by The Next Web (reliability 78/100). Altman noted that while AI can increase productivity, it also risks users ceasing to exercise their own critical reasoning, creativity, and memory. The full interview is available on YouTube and has garnered over 2 million views in its first week, reflecting public interest in the topic.

Why is it important?

Altman's warning is significant for several reasons. First, it comes from the leader of the company that has driven the massive adoption of generative AI with products like ChatGPT, which surpassed 200 million monthly active users in 2024. That the CEO himself acknowledges a risk inherent to his product adds credibility and urgency to the debate. The concept of cognitive atrophy is not new: psychologists like Daniel Kahneman have studied how reliance on mental shortcuts can reduce deep reasoning ability. However, the speed of AI integration is unprecedented. According to a 2024 Stanford University study, 40% of office workers in the US already use generative AI at least once a week, and that figure doubles among those under 30. Cognitive atrophy could have long-term effects on innovation and society's intellectual resilience. As neuroscientist Michael Merzenich noted in 2023, “the brain is like a muscle: if you don't use it, it atrophies,” and AI could accelerate that process by outsourcing key cognitive functions.

Consequences for businesses, users, and the market

For users

The most immediate risk is the loss of fundamental skills: memory, critical reasoning, creativity, and judgment. A 2024 Harvard University study found that students who used ChatGPT to write essays showed a 15% decline in their independent argumentation ability after three months of continuous use. Additionally, reliance on AI for everyday tasks like summarizing documents or planning trips could reduce neural plasticity. In the long term, this could translate into a diminished capacity to innovate and adapt to unforeseen situations, especially in younger generations who grow up with AI as a default tool.

For businesses

Organizations that adopt AI without a training and oversight strategy could see a decline in the quality of intellectual work. A 2024 McKinsey report estimates that 30% of office tasks could be automated with AI, but warns that lack of human oversight could lead to systemic errors. For example, in 2024, a New York law firm was sanctioned for submitting AI-generated legal documents containing fabricated citations. Companies that promote balanced use, combining AI with critical thinking training, could gain a competitive advantage. A notable case is IBM, which in 2023 launched an “AI literacy” program for its employees, requiring them to verify at least two external sources before accepting any AI-generated result.

For the tech market

Altman's warning could influence future product design. OpenAI has already introduced features like “verification reminders” in ChatGPT Enterprise, which ask users to confirm data accuracy before sending an email. Other companies like Google and Microsoft could follow suit, integrating “cognitive training” modes that challenge users to solve problems without assistance. Additionally, regulation could accelerate: in the European Union, the AI Act (effective 2025) requires high-risk systems to include human oversight mechanisms. Cognitive atrophy could be classified as a psychosocial risk, forcing companies to implement safeguards.

Comparison with past events

The debate echoes warnings about the “Google effect,” coined by Betsy Sparrow in 2011, which showed that people tend to forget information they know they can easily find online. There are also parallels with the loss of navigation skills due to GPS, documented in a 2017 University College London study, which showed that GPS users have reduced activity in the hippocampus, the brain region responsible for spatial memory. However, generative AI is qualitatively different: it not only stores or retrieves information but generates new content, from emails to computer code. This outsources not just memory but also the creative and decision-making process. As philosopher Nicholas Carr noted in his book The Shallows (2010), technology shapes our thinking; AI could accelerate that transformation toward shallower thinking. Altman himself compared the risk to that of television in the 1950s, which was criticized for fostering intellectual passivity, though AI has far greater potential to automate complex mental processes.

What should readers know?

It is crucial to maintain balance. Use AI as a support tool, not a substitute for your own thinking. Altman recommended in the interview “practicing discomfort” by solving problems without AI from time to time. Always verify generated results, especially in professional or academic contexts. Foster critical skills in children and young people through activities like debates, creative writing, and problem-solving without technological aid. Altman did not propose concrete solutions, but his warning underscores the need for social and institutional debate. For example, some schools in Finland have already implemented policies banning AI use in certain tasks to preserve critical thinking. In the workplace, leaders should design workflows that integrate AI without eliminating human reflection. Cognitive atrophy is not inevitable, but it requires awareness and collective action to prevent technological convenience from undermining our intellectual autonomy.