they’re reinforcement learning models, obviously

the RL guys from many decades ago were right all along