Scalable Environments Drive Generalizable Agents

In three linesPosition paper arguing that generalizable agents require environment scaling—expanding the distribution of executable rule-sets agents interact with—beyond trajectory or task scaling within fixed benchmarks. Proposes unified taxonomy separating trajectory, task, and environment scaling; synthesizes construction paradigms (programmatic generators vs generative world models) for scalable environments.

Summary generated by Claude — human-verified