AI researchers clash over how training data shapes neural networks, ends in a block
JacquesThibs · x · 2026-09-09
A debate over how training data works: one side argues it either contributes to a network's world model or reinforces token paths, reducing search and enabling distillation; the other disputes the mechanism. The exchange ended with a block.
More from Fun
- Biologist has GPT-6 Astra design a custom pipette from scratch for 3D printing — DeryaTR_ · 2026-09-09
- Why you should catch up on Michael Levin's takes on intelligence — cephaloform · 2026-09-09
- AI circle meme: asking Sam Altman to fund the cat-ears peptide next — QuintinPope5 · 2026-09-09
- Devs turn a retired Meta Portal into a baby-care voice logger using gpt-realtime API — SIGKITTEN · 2026-09-09
- Everyone's still warming up to one agent; OpenAI is unleashing 10,000 on math nerds — msg · 2026-09-09
- Pedro Domingos: LLMs still can't add two numbers of arbitrary length — pmddomingos · 2026-09-09