#Automated Evaluation

AgentHorizon Tests AI Genuine Models That Evaluate Long Tasks

New research queries the reliability of automatic evaluators in complex multi-app tasks of computer-user agents