{"data":{"event":{"id":"80b1fda4-30a0-4f63-b2a0-b1b4986fcd64","slug":"gpt5-pro-nesterov-point-convergence","title":"Ryu and Jang, using GPT-5 Pro, prove point convergence of Nesterov accelerated gradient","short_summary":"In October 2025 UCLA mathematicians Ernest Ryu and Uijeong Jang resolved the long-open question of whether Nesterov’s 1983 accelerated gradient method converges in iterates, in a collaboration they describe as heavily assisted by GPT-5 Pro. The human authors wrote and checked the final proof.","full_description":"Nesterov acceleration has been a workhorse of convex optimization since 1983, but point convergence of the iterates (as opposed to function-value rates) had remained open. Ryu and Jang announced continuous-time convergence on 21 October 2025 and discrete NAG on 24 October, then posted arXiv:2510.23513. They say GPT-5 Pro generated many arguments, most incorrect; Ryu filtered them and recognized a structural rearrangement that became the backbone of a human-written proof. OpenAI later covered the collaboration. This is material acceleration by a named GPT-5 Pro release, not autonomous authorship and not GPT-5.6 Sol.\n\nUpdate (2025-10-28): Review of the Anthropic Summer 2025 Pilot Sabotage Risk Report October 28, 2025 External review from METR of Anthropic's Summer 2025 Sabotage Risk Report Read more. Review of the Anthropic Summer 2025 Pilot Sabotage Risk Report October 28, 2025 External review from METR of Anthropic's Summer 2025 Sabotage Risk Report Read more\n\nUpdate (2025-10-23): Summary of our gpt-oss methodology review October 23, 2025 Details on external recommendations from METR for gpt-oss Preparedness experiments and follow-up from OpenAI. Read more. Summary of our gpt-oss methodology review October 23, 2025 Details on external recommendations from METR for gpt-oss Preparedness experiments and follow-up from OpenAI. Read more","ledger_type":"benefit","primary_domain_id":"23d935bb-b857-4b2b-9471-d568bb848a66","event_status":"provisional","event_date":"2025-10-24T00:00:00.000Z","discovery_date":"2025-10-24T00:00:00.000Z","first_published_date":"2025-10-24T00:00:00.000Z","last_reviewed_date":"2025-10-23T00:00:00.000Z","geographic_scope":"United States","affected_population":"Optimization theory and machine-learning research communities","base_impact_tier":3,"base_score":"10.00","attribution_multiplier":"0.4500","evidence_multiplier":"0.5000","realization_multiplier":"0.6000","durability_multiplier":"0.7500","current_event_score":"1.012500","confidence_level":"medium","score_explanation":"Resolving NAG point convergence is a major optimization theorem (tier 3). Humans led and wrote the proof (AI attribution 0.45). Evidence is an arXiv paper plus an OpenAI writeup (0.45). Realization is a public complete argument (0.60). Follow-up ingest: independent source raised evidence; added a follow-up source. Follow-up ingest: added a follow-up source.","methodology_version_id":"d7881163-fb23-4e71-8625-bac1a8662c0f","original_methodology_version_id":"d7881163-fb23-4e71-8625-bac1a8662c0f","published_at":"2025-10-24T00:00:00.000Z","created_at":"2026-09-13T22:56:05.431Z","updated_at":"2026-09-17T03:00:52.788Z","flags":[],"domain_name":"Mathematics","domain_slug":"mathematics","methodology_version":"0.1"},"contributions":[{"id":"8282aed4-844c-4e64-91eb-2b3d212a9d97","event_id":"80b1fda4-30a0-4f63-b2a0-b1b4986fcd64","model_id":"6ab16a1b-9da8-47af-b00b-3b42259e1279","role_description":"Generated candidate arguments and a structural idea that Ryu developed into the final NAG point-convergence proof.","attribution_multiplier":"0.4500","credit_share":"1.0000","contribution_score":"1.012500","attribution_rationale":"The arXiv paper states the work was conducted with GPT-5 Pro; Ryu describes filtering mostly incorrect arguments and writing the proof himself.","attribution_confidence":"medium","first_used_date":"2025-10-21T00:00:00.000Z","model_version_if_known":null,"review_status":"approved","created_at":"2026-09-13T22:56:05.431Z","model_slug":"gpt-5-pro","model_name":"GPT-5 Pro","identity_class":"named_release","is_internal":false,"family_name":"GPT","family_slug":"openai-gpt","organization_name":"OpenAI","organization_slug":"openai"}],"sources":[{"id":"fdcd4134-6391-467a-8ba3-8dbf4d8c0973","event_id":"80b1fda4-30a0-4f63-b2a0-b1b4986fcd64","url":"https://arxiv.org/abs/2510.23513","canonical_url":"https://arxiv.org/abs/2510.23513","source_type":"preprint","publisher":"arXiv","author":null,"publication_date":"2025-10-27T00:00:00.000Z","retrieved_at":"2026-09-13T22:56:05.431Z","title":"Point Convergence of Nesterov’s Accelerated Gradient Method: An AI-Assisted Proof","excerpt":null,"content_hash":null,"source_reliability_class":"medium","is_primary_source":true,"is_independent":true,"is_peer_reviewed":false,"archived_url":null,"created_at":"2026-09-13T22:56:05.431Z"},{"id":"56cee307-0083-4b8b-b0a5-72336a6fe859","event_id":"80b1fda4-30a0-4f63-b2a0-b1b4986fcd64","url":"https://metr.org/blog/2025-10-23-gpt-oss-methodology-review/","canonical_url":"https://metr.org/blog/2025-10-23-gpt-oss-methodology-review/","source_type":"other","publisher":"METR Blog","author":null,"publication_date":"2025-10-23T00:00:00.000Z","retrieved_at":"2026-09-17T03:00:52.781Z","title":"Summary of our gpt-oss methodology review October 23, 2025 Details on external recommendations from METR for gpt-oss Preparedness experiments and follow-up from OpenAI. Read more","excerpt":"Summary of our gpt-oss methodology review October 23, 2025 Details on external recommendations from METR for gpt-oss Preparedness experiments and follow-up from OpenAI. Read more","content_hash":null,"source_reliability_class":"medium","is_primary_source":false,"is_independent":true,"is_peer_reviewed":false,"archived_url":null,"created_at":"2026-09-17T03:00:52.781Z"},{"id":"50760236-7de9-471d-a6b0-48ba65a71a74","event_id":"80b1fda4-30a0-4f63-b2a0-b1b4986fcd64","url":"https://metr.org/blog/2025-10-28-sabotage-report-review/","canonical_url":"https://metr.org/blog/2025-10-28-sabotage-report-review/","source_type":"other","publisher":"METR Blog","author":null,"publication_date":"2025-10-28T00:00:00.000Z","retrieved_at":"2026-09-17T03:00:52.700Z","title":"Review of the Anthropic Summer 2025 Pilot Sabotage Risk Report October 28, 2025 External review from METR of Anthropic's Summer 2025 Sabotage Risk Report Read more","excerpt":"Review of the Anthropic Summer 2025 Pilot Sabotage Risk Report October 28, 2025 External review from METR of Anthropic's Summer 2025 Sabotage Risk Report Read more","content_hash":null,"source_reliability_class":"medium","is_primary_source":false,"is_independent":true,"is_peer_reviewed":false,"archived_url":null,"created_at":"2026-09-17T03:00:52.700Z"},{"id":"55c31d39-b5b4-43f6-899b-3422176144cd","event_id":"80b1fda4-30a0-4f63-b2a0-b1b4986fcd64","url":"https://openai.com/index/gpt-5-mathematical-discovery/","canonical_url":"https://openai.com/index/gpt-5-mathematical-discovery/","source_type":"developer_announcement","publisher":"OpenAI","author":null,"publication_date":"2025-11-24T00:00:00.000Z","retrieved_at":"2026-09-13T22:56:05.431Z","title":"How GPT-5 helped mathematician Ernest Ryu solve a 40-year-old open problem","excerpt":null,"content_hash":null,"source_reliability_class":"low","is_primary_source":false,"is_independent":false,"is_peer_reviewed":false,"archived_url":null,"created_at":"2026-09-13T22:56:05.431Z"}],"claims":[{"id":"c696fcfa-241d-4011-a906-4dde8b1dd55d","event_id":"80b1fda4-30a0-4f63-b2a0-b1b4986fcd64","claim_text":"Ryu and Jang proved point convergence of Nesterov accelerated gradient, in a collaboration they describe as heavily assisted by GPT-5 Pro.","claim_type":"outcome","claim_status":"supported","confidence_score":null,"created_at":"2026-09-13T22:56:05.431Z","updated_at":"2026-09-13T22:56:05.431Z"},{"id":"3ce86402-5239-47ac-91ec-5b3658960e6c","event_id":"80b1fda4-30a0-4f63-b2a0-b1b4986fcd64","claim_text":"GPT-5 Pro autonomously authored the published proof without human filtering.","claim_type":"attribution","claim_status":"disputed","confidence_score":null,"created_at":"2026-09-13T22:56:05.431Z","updated_at":"2026-09-13T22:56:05.431Z"}],"revisions":[{"id":"a81e045f-332e-443b-a781-0c1d14026768","event_id":"80b1fda4-30a0-4f63-b2a0-b1b4986fcd64","model_id":null,"previous_score":"0.000000","new_score":"0.911300","previous_factors":{},"new_factors":{"evidence":0.45,"base_score":10,"durability":0.75,"attribution":0.45,"realization":0.6},"change_reason":"Imported events under methodology 0.1.","trigger_type":"editorial_import","trigger_source_ids":null,"reviewer_id":null,"review_status":"published","created_at":"2026-09-13T22:56:05.431Z"},{"id":"23c6bc96-a3df-4af9-bcb6-8a78ec3727ed","event_id":"80b1fda4-30a0-4f63-b2a0-b1b4986fcd64","model_id":null,"previous_score":"0.911300","new_score":"1.012500","previous_factors":{"tier":3,"evidence":0.45,"base_score":10,"durability":0.75,"attribution":0.45,"realization":0.6},"new_factors":{"tier":3,"models":["6ab16a1b-9da8-47af-b00b-3b42259e1279"],"evidence":0.5,"base_score":10,"durability":0.75,"attribution":0.45,"realization":0.6},"change_reason":"Ingest update: independent source raised evidence; added a follow-up source.","trigger_type":"ingest_update","trigger_source_ids":["50760236-7de9-471d-a6b0-48ba65a71a74"],"reviewer_id":null,"review_status":"published","created_at":"2026-09-17T03:00:52.718Z"},{"id":"92661691-c40c-4ce7-837b-b3222ee0a166","event_id":"80b1fda4-30a0-4f63-b2a0-b1b4986fcd64","model_id":null,"previous_score":"1.012500","new_score":"1.012500","previous_factors":{"tier":3,"evidence":0.5,"base_score":10,"durability":0.75,"attribution":0.45,"realization":0.6},"new_factors":{"tier":3,"models":["6ab16a1b-9da8-47af-b00b-3b42259e1279"],"evidence":0.5,"base_score":10,"durability":0.75,"attribution":0.45,"realization":0.6},"change_reason":"Ingest update: added a follow-up source.","trigger_type":"ingest_update","trigger_source_ids":["56cee307-0083-4b8b-b0a5-72336a6fe859"],"reviewer_id":null,"review_status":"published","created_at":"2026-09-17T03:00:52.796Z"}],"secondary":[{"name":"Computer Science","slug":"computer-science"}]},"methodology_version":"0.1"}