Feedback-Loop Amplification
A model magnifies a slight human bias; we absorb the magnified version and feed it back, larger each pass.
ExampleAsk for pictures of a manager, see the same face over and over, and slowly start expecting that face.
What to doNotice when a model’s answer is just your own view, stated louder. Seek sources the model did not shape.
Paste this into the chatI think the following is true. Before agreeing, give me the best evidence against it, and tell me how confident I should be if I remove my own framing.