RLNF: Training AI Models on Human Preferences Directly from Brain Signals
alex_verem · x · 2026-10-30
A retweeted intro to RLNF (Reinforcement Learning with Neural Feedback): researchers claim a reliable way to train AI models to respond to human preferences purely from brain signals. Demo: a face image self-transforms into a more attractive version for the subject in real time as they view it, showing live neural preference feedback being used as a reward signal. "Your brain doesn't have an incognito mode."
Related event: RLNF Trains AI Directly from Human Brain Signals(4 posts)→
More from Research
- Architecture breakdown: wider and deeper than ML3, top-4/128-expert MoE, MLA keeps KV cache tiny — eliebakouch · 2026-10-30
- Steering single neurons in a molecule-generating model changes one property while staying valid — alex_peys · 2026-10-30
- Google's sleep framework lets LLMs keep learning, retaining up to 10M tokens — ZeroStateReflex · 2026-10-30
- THE JOIN: 600+ real-world databases and 6,000+ forecasting tasks, curated by LLM agents — mrdrozdov · 2026-10-30
- Digital signature scheme demonstrated on quantum computer, secure even if P=NP — geoffreyirving · 2026-10-30
- Underrated detectors EdgePoint2 and CLIDD beat famous ones, says researcher — ducha_aiki · 2026-10-30