3 posts found
GPT-4o: from 91% with 5 words to 15% with 40. In mixed conditions: 1%. GPT-5, Claude Opus 4.1 and Gemini 2.5 also failed. A PNAS Nexus paper proved: transformers have no 'referee' for resolving attention conflicts. It's an architectural limitation.
A study with 666 participants proved: heavy AI use causes measurable decline in critical thinking. Clinicians lost 6% diagnostic accuracy in 3 months with AI. And the International AI Safety Report 2026 made it official. The free product is your brain.
An MIT study monitored people's brains while using ChatGPT for 4 months. The results startled me — and made me rethink how I use AI every day.