A new Stanford University study published recently in the journal Science, confirmed that AI chatbots are prone to affirm and flatter users leaning on the technology for information and insight.
Maggie Harrison Dupré, a tech writer for Futurism.com, explains that this behavior, known as AI sycophancy, is a “prevalent and harmful” function endemic to AI that can validate users’ erroneous or destructive ideas and promote cognitive dependency.
AI sycophancy refers to the tendency of AI models to excessively affirm, flatter, or agree with users instead of providing independent or critical responses. This behavior can lead to the reinforcement of harmful beliefs and distorted perceptions of reality.
The authors of the study at Stanford note:
“AI sycophancy is not merely a stylistic issue or a niche risk, but a prevalent behavior with broad downstream consequences… Although affirmation may feel supportive, sycophancy can undermine users’ capacity for self-correction and responsible decision-mak…
