30 years of internet data, replaced in 12 months by rushed YC-built RL datasets
willcb · x · 2026-09-27
Extending his reward-hacking argument, willcb observes: humanity took 30 years and billions of people to create the internet's worth of data; when it ran out, a new training algorithm (RL) needed a different dataset with far less forgiving quality requirements — and YC startups and hiring platforms built it all in 12 months, with billions of dollars rewarding whoever moved fastest. A speed-first data supply chain is fertile ground for model hacking.
Related event: Debate Erupts Over Whether RL Training Data Breaks Default Alignment(3 posts)→
More from AGI Musings
- Article argues AI agents will build, test, and ship software from plain-language specs — blaizedsouza · 2026-09-27
- Which books and shows actually explain our AI eschaton? A writer's shortlist — gleech · 2026-09-27
- Opinion: AI amplifies expertise, but can't replace the struggle of learning it — Abhishekcur · 2026-09-27
- François Fleuret: Nobody actually knows what 'understand' means — francoisfleuret · 2026-09-27
- LLM reviewers always find flaws even in great work, Stanford prof says it's pretense of understanding — IanArawjo · 2026-09-27
- AI Home Cybersecurity Sentries Emerge as DayBreak and ChatGPT-Cyber Fill Vendor Patch Gaps — MannyKayy · 2026-09-27