PNAS paper: kids explore farther than RL agents, driven by empowerment not info gain
_akpiper · x · 2026-11-06
A new PNAS paper by Eliza Kosoy, Pieter Abbeel, and Alison Gopnik places kids (8-12), adults, and intrinsically motivated RL agents in the same open-ended environment to study what drives exploration without instructions.
Key findings:
- Humans explored farther despite far less time
- Human exploration correlated more consistently with novelty and empowerment (gaining control over the environment) than with information gain
- Across humans and agents, novelty-driven exploration plateaued while empowerment kept increasing
The authors suggest the next frontier in agentic AI may be understanding curiosity itself, not just better reasoning or reward functions.
More from Research
- SEABench shows self-evolving agents can cause harm by rewriting their own harness, no adversary needed — nandofioretto · 2026-11-06
- DFlash 2 parallel drafting lifts speculative decoding acceptance 21% for 1.3% latency cost — songhan_mit · 2026-11-06
- OpenTac Brings HDR Readout to Open Tactile Sensors, Sensing Light and Heavy Touch at Once — _krishna_murthy · 2026-11-06
- METR publishes independent assessment of Anthropic's AI R&D acceleration, evidence heavily redacted — rohanpaul_ai · 2026-11-06
- Model value as conditional entropy removed: why rising world entropy hurts model companies — xiaosun86 · 2026-11-06
- The most valuable secrets in AI are no-go theorems, not new tricks — fkasummer · 2026-11-06