Reinforcement Learning Requires Human-in-the-Loop Framing and Approaches

Reinforcement learning (RL) is typically framed as a machine paradigm where agents learn to act autonomously in complex environments. This paper argues instead that RL fundamentally human the loop (HitL). The reward functions (and other components) of Markov decision process are defined by humans. decisions tackle certain problem, and deploy learned solution, taken Humans can also play critical role providing information agent throughout its life cycle better succeed at problem question. We end highlighting set HitL research questions, which, if ignored, could cause fail live up full potential.