16:46
2026-09-03
provenbrief.com
ai-agents
Web agents' world models learn to predict, not to choose: a rebuilt training objective lifts task success
A paper posted 2 September by a team including Kelvin Li, Leonid Karlinsky, Rogerio Feris, Trevor Darrell and Roei Herzig argues that LLM web agent pipelines train world models with supervised next-st…