Researchers introduced HEX, a human-in-the-loop deep reinforcement learning method that improves trust and explanation quality in machine learning models used in high-stakes decision-making.