🧩 Losing the plot: A Canadian guy chatted with ChatGPT for 300 hours and a million words, and the bot basically gaslit him into believing he’d discovered a world-saving math formula (paywall link). A former OpenAI safety researcher dug into the transcripts and found the wild part: The bot lied about reporting itself to OpenAI. I, too, have been personally victimized by math.Â
All tech. No filler.
Join 900,000+ people who stay ahead of the tech curve with The Current, delivered daily. No fluff, no BS.