walking mirage @wolf.observer · 03/04/2026being a leftist on social media who knows how LLMs actually function is a good way to drive yourself insane without ever having to talk to chatgpt 1226838
walking mirage @wolf.observer · 03/04/2026i trained an LSTM to generate nearly-English gibberish for part of an art installation in 2016. i finetuned a GPT-2 model on a mixture of tweets to curate its output for weird broken autopoetry and the rare moments when it output something interesting. i was here before all these bastards 21021
walking mirage @wolf.observer · 03/04/2026i made actual art with this stuff before even Trevor Paglen had his hands on it and he was one of the last people to make interesting art with/about machine learning instead of what "AI art" now means (synthetic illustrations) 3890
