How Good Are Human Trained AI Models for Training Humans?

Interacting with large language models