@shituationist@kolektiva.social
Post #2065481
2026-05-04 00:50 UTC
Seems like no one gets chatbot psychosis from Gemini, just Claude, less often Chatgpt. What's up with that? There's a specific flavor to Anthropic-induced chatbot psychosis too. Openai-induced psychosis is always more "apollonian", in a way, more in the "guys, I think I just 'cracked physics'" sort of way. The Anthropic variant is always more dilletantish, centrist, kind of like "guys, I think the chatbot has become 'real', and it's also my 'friend'".
Replies (1)
-
@dogiedog64@app.wafrn.net 2026-05-04 01:05
@shituationist@kolektiva.social I don't hang around with enough botbrains to notice that kind of madness manifesting, but that sounds to me like the differing levels of sycophancy, and the emphasis of the training data, causing different levels of damage. Obviously none of it is good, but that's definitely an interesting observation.