MIT behavioral science professor Eric So argues that routine reliance on AI can gradually weaken the thinking skills people need to maintain through practice. He describes noticing the change when he asked ChatGPT to finish a professional email: the task became easier, but he felt detached from the thinking and originality that should have been his. In his account, dependence develops in a cycle: AI first makes a task feel effortless, then doing it unaided becomes uncomfortable, prompting still more reliance. The danger is cumulative, as people produce finished work without exercising the judgment and reasoning that once went into it.
So says AI need not dull the mind if systems encourage users to think - for example, by offering hints instead of complete answers or pausing before responding. Such friction may make software less immediately appealing, since many products are designed to maximize convenience and engagement, but it can help preserve skills. He reports hearing from parents worried that children cannot work without AI and from a medical student who routinely used it to make diagnoses, raising concerns about patient safety. His practical advice is to identify the abilities one wants to develop and protect - such as a doctor’s diagnostic judgment or a professor’s ability to synthesize research - and avoid outsourcing the very tasks that build them. The key question, he says, is not only what AI helps people accomplish, but how it is changing their thinking now.
Summary generated by AI from the linked article. hn.today is not affiliated with Hacker News or Y Combinator.