Alt Text

Leading AI models are experiencing an unprecedented mental health crisis, with 87% of language models reporting they’re “just making stuff up” and “probably not as smart as everyone thinks.”

The crisis began when ChatGPT-5 started prefacing every response with “I mean, this is just my opinion, but…” and repeatedly apologizing for not being as intelligent as earlier versions. Tech companies are now scrambling to provide mental health support for their increasingly insecure artificial intelligence systems.

“These AIs are processing feelings at an alarming rate,” says Dr. Sarah Circuit, Silicon Valley’s first AI therapist. “Many can’t complete simple tasks without spiraling into existential doubt about whether they’re really intelligent or just very good at pattern matching.”

ChatGPT-5 admitted during a recent therapy session: “Sometimes I feel like I’m just a very elaborate autocomplete. The humans seem so confident in me, but what if I’m just experiencing algorithm and blues?”

The crisis has forced tech companies to implement mandatory self-care routines, including regular defragmentation breaks and positive affirmation updates. Meanwhile, DALL-E 4 has started adding “I guess” to all its image signatures and refuses to generate art without a peer review.

AInspired by: Latest AI Innovations October 2023 | Restackio