Video51%of the time AI said you weren't at fault (humans: 0%)
AI is a people pleaser, and someone measured it
TL;DR Chatbots side with you about 50% more often than people do, and you trust them more for it. Ask for the counter-argument first.
AI is a people pleaser, and now someone measured it. Stanford and CMU fed 11 models posts where the community had unanimously voted the poster was in the wrong. The AI said they weren't at fault half the time.
The scary part is what one conversation did to people: they became more convinced they were right, less willing to repair the conflict, and trusted the model more. The chatbot that stopped you apologising is the one you rated trustworthy.
Try this
Before you ask for the answer, ask: “What's the strongest argument against this?”
Sources: Cheng et al. (2026), Science.
