Breaking the Supervised Learning Curse: Models Must Learn 'In the Wild'
tensorqt · x · 2026-08-09
Responding to John Schulman's point about models struggling with controversial topics from first principles, the commentary argues this is the inherent curse of supervised (including self-supervised) training. While highly effective, this paradigm is information-bound by its pre-text tasks. To break this barrier, models must learn 'in the wild', accepting all the implied risks that come with it.
More from AGI Musings
- AI Lowers the Barrier to Building, Not to Understanding Industries — iamKierraD · 2026-08-09
- AI Compute Costs: UK Datacentre Expansion Sparks Water and Power Crises — nordicinst · 2026-08-09
- Time Magazine Starts Serving Ads Directly to AI Agents — 233C · 2026-08-09
- Fei-Fei Li on Spatial Intelligence: Data is Harder Than Models — FinanceYF5 · 2026-08-09
- AI-generated lawsuits flood UK employment courts, backlog jumps 55% — The Decoder · 2026-08-09
- Opinion: Beyond Scaling, Model Improvements Lack Generalization, Reducing to Benchmark Climbing — LChoshen · 2026-08-09