Unreliable Systems

Percy and Lukas discuss the unreliability of current agent-based systems in machine learning tasks. While improvements are anticipated in the future, the need for human verification and guidance is emphasized to make these systems more useful in practical applications.