{"type":"post","id":"llms-are-not-superhuman-at-reading-and-writing-c6afd7","slug":"llms-are-not-superhuman-at-reading-and-writing-c6afd7","title":"LLMs are not superhuman at reading and writing neural network weights","description":"Sean Goedecke argues that GPT-6 Astra’s direct reading and editing of neural-network weights is impressive but not evidence of superhuman AI research.","retrieval_nugget":"Sean Goedecke argues that GPT-6 Astra’s direct reading and editing of neural-network weights is impressive but not evidence of superhuman AI research. The clearest result, he says, fixed a specific long-context looping failure in a 0.8B Qwen model. For builders, the distinction matters: a targeted parameter patch is not the same as general self-improvement.","published_at":"2026-11-03","updated_at":"2026-11-06","record_date":"2026-11-03","date_kind":"published_at","topics":["models"],"entities":["Sean Goedecke","GPT-6 Astra","Qwen","Qwen3.5-0.8B"],"related_patterns":["scientific-agents-need-artifact-evidence"],"novelty_label":"new_angle","what_changed":"The author argues the reported weight-reading and model-editing feats are not superhuman, explaining that a capable human ML researcher could perform them.","prior_refs":[],"source_url":"https://www.seangoedecke.com/llms-are-not-superhuman-at-reading-and-writing-neural-network-weights","source_urls":["https://www.seangoedecke.com/llms-are-not-superhuman-at-reading-and-writing-neural-network-weights"],"schema_version":"newruntime-agent-readable-v0.2","stable_id":"post:llms-are-not-superhuman-at-reading-and-writing-c6afd7","status":"published","visuals":[],"editor":{"role":"editor","name":"Andrey Reshetnikov","profile":"https://newruntime.com/owner-profile.md","profile_json":"https://newruntime.com/owner-profile.json","same_as":["https://www.linkedin.com/in/reshetnikov1/","https://t.me/lapetuse"]},"routes":{"html":"https://newruntime.com/posts/llms-are-not-superhuman-at-reading-and-writing-c6afd7/","markdown":"https://newruntime.com/posts/llms-are-not-superhuman-at-reading-and-writing-c6afd7.md","json":"https://newruntime.com/posts/llms-are-not-superhuman-at-reading-and-writing-c6afd7.json"},"markdown_url":"https://newruntime.com/posts/llms-are-not-superhuman-at-reading-and-writing-c6afd7.md","next_reads":[{"type":"topic","path":"/topics/models/","url":"https://newruntime.com/topics/models/","title":"Models","reason":"Explore the models topic hub.","media_type":"text/html","markdown_url":"https://newruntime.com/topics/models.md","json_url":"https://newruntime.com/topics/models.json"},{"type":"collection","path":"/posts/","url":"https://newruntime.com/posts/","title":"Field Notes","reason":"Open the latest editorial analysis.","media_type":"text/html","json_url":"https://newruntime.com/posts.json"},{"type":"collection","path":"/sources/","url":"https://newruntime.com/sources/","title":"Sources","reason":"See the original sources this archive cites.","media_type":"text/html"}]}