ResearchNovember 04, 2026
-24
š§
AI Now Writes Neural Network Brains Directly
LLMs now read and edit neural network weights directly, no gradient training needed.
#LLM#Weight Space Programming#Mechanistic Interpretability#Research
š„ What happened
Frontier models like GPT-6 Astra now edit neural network weights directly via text ā no backprop. Astra pushed Qwen3.5-0.8B from 7.9% to 48.1% on MATH-500 by rewriting 131,072 weights. Multiple independent replications confirm the effect, including a laptop run by Simon Willison.
š” Why it matters
Most of the gain is repair: 576 of 579 fixed answers were previously stuck in a 32K-token loop. On held-out problems, accuracy jumped from 10.6% to 60.4% ā existing reasoning that can finally finish. Critics call it bug-fixing, not recursive self-improvement.
ā” Our take
Fitting weights directly against a benchmark is overfitting with extra steps. The real story: coding agents suddenly wield mechanistic-interpretability tools effectively ā that's the actual paradigm shift.
The title, summary and analysis of this item were produced automatically by an AI system and have not been editorially reviewed. They may contain errors, bias or omissions ā when in doubt, read the linked original source.
