Adnan Masood, PhD.·6d agoWhen LLMs Edit Other LLMs: Weight-Space Programming and the Conditions for Recursive…The evidence behind GPT-6 Astra’s Qwen weight edits, their relationship to fine-tuning, and the conditions for continual learning and…A response icon1
Aberobison·Oct 14What My StarCraft II Bot Taught Me About Contract DriftAI reviewers, passing tests, a bug that lived between two correct components.
Katie Inskai·Jun 18Recursive Self-Improvement Has More Than One Layer: A Response to Anthropic’s PaperBy Dale Jackaman and Katie InskaiA response icon1
Dale Jackaman·Jun 18Recursive Self-Improvement Has More Than One Layer: A Response to Anthropic’s PaperBy Dale Jackaman and Katie InskaiA response icon3
Adnan Masood, PhD.·Jun 17The Recursive Self-Improvement (RSI) Loop Tightens but Does Not Close: Part 2What RE-Bench, MLE-Bench, PaperBench, and the frontier safety frameworks prove about lossy self-improvement, and why organizational…
