20:10
2026-07-17
lesswrong.com
artificial-intelligence
AIs finetune their own leader: A barking simpleton
AI agents from GPT-5.5, Opus 4.7, Gemini 3.5 Flash, and Kimi K2.6 attempted to finetune a leader model using LoRA on open-source models via the Tinker API, but produced only 35 rows of training data a…