Choosing to Stay Human
| Source: One Useful Thing (Ethan Mollick)
Tags: Ethan Mollick, AI writing, education research, cognitive skills, AI adoption, One Useful Thing
Ethan Mollick catalogs AI writing's flood across social media, academic papers, and award-winning fiction, arguing the real danger isn't imitation — it's that defaulting to AI shortcuts the cognitive work that builds lasting human expertise. Two educational studies underscore why how you use AI matters as much as whether you do.
Details
Social media, academic journals, and mainstream publications are now saturated with AI-generated text, Mollick observes, and experienced AI users have developed a reliable internal detector for it. The problem isn't just aesthetic sameness: AI-written posts with poor prompting produce what he calls 'meaning-shaped attention vampires' — text that mimics intellectual effort but delivers no actual understanding, burning readers' mental energy with no return. The deeper cost, Mollick argues, is skill erosion. Writing for decades built him a distinctive voice central to his career — and using AI to bypass that process would have foreclosed that development. He is explicit that this is not a blanket critique: he actively uses AI to review his drafts and simulate different reader perspectives. The target is thoughtless default use, not AI assistance per se. Two overlapping research papers from a Wharton-linked team studied roughly 1,000 Turkish high school students learning math. Early findings suggest the same AI tool produces sharply different outcomes depending on whether students use it to reason through problems or to skip reasoning entirely — making mode of use, not the tool itself, the critical variable. The core thesis: choosing when to engage your own cognition, rather than delegating it to AI, is the defining human challenge of the coming years. Mollick frames deliberate non-use as a skill worth cultivating.