Conversational AI that returns ready answers changes how people practice cognition: users stop training evaluative skills, critics and experts are displaced by plausibly fluent but shallow outputs, and social incentives favor quick AI answers over slower scrutiny. Over time this produces measurable declines in public reasoning, increases in confidence without competence, and a feedback loop where AI content lowers the quality of human discourse.
— If true, it implies widespread deployment of chatty AI will reshape education, journalism, civic debate, and regulatory priorities by degrading collective epistemic capacity.
Erik Hoel
2026.03.05
100% relevant
The article’s central claim that 'six years of AI and the world got stupider,' plus its descriptions of chat interfaces, METR‑driven investment hype, and flat reliability despite benchmark gains, exemplify this phenomenon.
← Back to All Ideas