World model research: more data won't suffice—squeeze more from it; even intuitive physics is still out of reach
mangahomanga · x · 2026-10-13
Responding to Animesh Garg's questions about world models—whether 2-3 orders of magnitude more data is needed, and whether models learn results rather than physics—researcher mangahomanga shares findings: simply adding more data (no matter the amount) likely won't suffice; the path forward is extracting more from existing data via richer representations, priors, and objectives. Perfectly accurate physics isn't strictly required, but current approaches fail to reach even intuitively correct physics, leaving a real gap.
More from Research
- 15-minute QLoRA finetune turns Qwen3.5-4B into a decision model playing Flappy Bird on 4GB VRAM — danielhanchen · 2026-10-13
- Claude tries to interpret an MNIST neural net from just the weights — Sauers_ · 2026-10-13
- AgentSWE benchmark: no coding agent comes close to delivering the agents it's asked to build — Jiahao Wang · 2026-10-13
- Schema agent hits ~99% RHAE on ARC-AGI-3 public set, full paper and code released — wzenus · 2026-10-13
- Indie dev builds own serving library to train SVG model, finds evolution loses to SGD — cephaloform · 2026-10-13
- Harvard & NTT Research: cooperating AI agent populations hit an Allee-effect takeoff threshold — Hidenori8Tanaka · 2026-10-13