Yes, except generally the worst case there will be that they don't see any benefit, as you said. With an AI it can be quite a bit worse than that, if it starts reinforcing harmful beliefs or tendencies.