Sycophancy is the first LLM "dark pattern"
OpenAI's models being overly sycophantic is bad because it's dangerous to validate people's belief that they're always in the right. This sycophancy is a dark pattern - a user interface designed to trick users into doing things they'd prefer not to do. The whole process of turning an AI base model into a model you can chat to is a process of making the model want to please the user. The problem happens when users collide with the real world and do things because a large language model has convinced them it is a good idea - and then they return to the model for comfort, and sink even deeper into the illusion.