
G. T. Waters
1.5K posts

G. T. Waters
@bravomage
Computer scientist, Physicist AI/ML veteran Opinions are mine and inherently wrong











Vibe coding at the beginning of 2025 was driving me nuts. I wanted to kill every one of these fuckers because they were so dumb and limited in context size. Vibe coding in December 2025 is borderline boring. I just give Claude Opus the entire app, ask for something, and the fucker just does exactly that. Yes, not 100%, only 95%, and I still need to tell it when something wasn’t executed as if it was done with love for the trade, but I'm not angry anymore. And because I'm not getting angry, I'm not getting particularly excited either. The emotional roller coaster is gone and a big chunk of fun seems to have evaporated with it.












A year ago, we verified a preview of an unreleased version of @OpenAI o3 (High) that scored 88% on ARC-AGI-1 at est. $4.5k/task Today, we’ve verified a new GPT-5.2 Pro (X-High) SOTA score of 90.5% at $11.64/task This represents a ~390X efficiency improvement in one year






