MIT Professor Eric so Warns of AI Dependency in New Book

In a Boston Globe interview, So described noticing his own thinking change and urged people to identify the skills they want to protect.

By James Okafor

October 22, 2026 · 1 min read

MIT behavioral science professor Eric So said in an interview published by the Boston Globe on October 23, 2026 that heavy use of artificial intelligence tools is changing how people think, a theme he explores in his new book, "The Collision: What AI Does to Us."

So said he first recognized the problem while asking ChatGPT to finish a sentence in an email to a colleague, and felt he had taken his own thought process out of the work. He described a cycle in which users begin relying on AI out of curiosity or employer pressure, lose practice at the underlying skill, grow uncomfortable performing it unaided, and then turn to AI more often.

So told the Globe that chatbots can be designed to prompt thinking rather than supply answers — by offering hints or inserting a delay before responding — though many systems instead minimize friction to drive engagement. He said he believes designs that keep users sharp can still succeed.

He recounted a conversation with a medical student who said consulting an AI bot had become their default when asked for a diagnosis. So said people should identify the "signature thinking skills" they want to retain over the next five to 10 years and decide how to use AI from there.

This story was first reported by The Boston Globe.