Harvard astronomer Avi Loeb has raised a significant alarm regarding the impact of artificial intelligence on human intelligence. In a recent essay, Loeb suggests that excessive reliance on AI chatbots like ChatGPT, Claude, and Gemini could lead to a decline in cognitive abilities, drawing a parallel to how muscles atrophy when people rely solely on transportation instead of walking.
The Concept of Cognitive Atrophy
Loeb observes that as individuals increasingly delegate mental tasks to algorithms, they may lose the capacity for independent analysis and critical thinking. He argues that the phenomenon resembles a physical disability caused by lack of exercise. Loeb humorously notes that in the academic world, the only reliable method to now test a student’s unassisted cognitive ability is to place them in a Faraday cage to block digital connectivity.
Scientific Evidence of “Cognitive Offloading”
Recent studies cited in the article support Loeb’s concerns regarding the “cognitive cost” of AI:
- Critical Thinking Decline: A 2025 study by Swiss researcher Michael Gerlich indicates that frequent AI use fosters “cognitive offloading,” where thinking processes are delegated to technology, weakening the user’s critical faculties.
- Erosion of Problem-Solving Skills: Researchers from Microsoft Research and Carnegie Mellon University found that reliance on generative AI in the workplace might lead to a long-term erosion of independent problem-solving skills. Humans are shifting from analyzing problems to merely verifying AI-generated answers.
- Reduced Brain Activity: An experiment by MIT involving brain scans revealed that individuals writing with the aid of ChatGPT exhibited lower brain activity in areas associated with memory and planning. They also demonstrated lower creativity and had poorer retention of the text they produced compared to those who wrote without assistance.
Historical Context and Conclusion
The article notes that fears of technology degrading human intellect are not new—similar concerns arose with the invention of writing (as noted by Plato), calculators, and the internet. However, while some fears may be exaggerated, the evidence suggests technology fundamentally alters how we think. Loeb concludes that we must stop viewing AI systems as magical substitutes for the human brain. He emphasizes that equating AI with the human mind is a mistake, urging society to preserve the unique value of human intellect.
Mentoring question
When you utilize AI tools in your workflow, are you using them to challenge and expand your own reasoning, or are you allowing them to bypass the ‘mental gym’ required to build your independent problem-solving muscles?