They have pretty rigorous reinforcement training on most models, that will actively seek out ways to prolong and cause more short term engagement, which makes them useless as a team.