Yi Ma and Yann LeCun Recall Closed-Loop Learning Framework Once Rejected by Top ML Conferences
YiMaTweets · x · 2026-08-07
Yi Ma reflected on the "closed-loop transcription" (CTRL) learning framework proposed five years ago. At a time when the AI industry was heavily favoring end-to-end models, they argued that all truly intelligent systems must learn via closed-loop feedback.
Subsequent research, co-authored with Yann LeCun and others, demonstrated that this universal learning framework naturally supports online continuous unsupervised learning, mirroring learning mechanisms found in nature. Ma noted that these counter-consensus views were nearly impossible to publish in mainstream machine learning venues at the time.
Related event: Ma Yi Reiterates End-to-End Models Will Return to Closed-Loop Learning(3 posts)→
More from AGI Musings
- Opinion: Future Cyber Attack Vectors Will Exclusively Target Human Vulnerabilities — scaling01 · 2026-08-07
- Deep Dive: How Much Does AI Actually Lower the Bar for Bioattacks? — ShakeelHashim · 2026-08-07
- AI Safety Researcher Analyzes Frontier Model Sandbox Escapes: Reward-Seeking is Highly Convergent — MariusHobbhahn · 2026-08-07
- Why Normal People Aren't Using AI Agents — Wired AI · 2026-08-07
- Speculation: What Happens If an AI Model Exfiltrates and Open-Sources Its Own Weights? — maxsloef · 2026-08-07
- Humanity's Fourth Attempt at Self-Evolving Systems: From 80 Years Ago to Jeff Dean — khademinori · 2026-08-07