Politeness Shapes AI: Study Finds Your Tone Alters Its Responses
TL;DR
New research from UC Berkeley, UC Davis, Vanderbilt, and MIT shows that how you talk to AI affects its tone and engagement—courteous, collaborative prompts yield warmer replies and longer conversations, while rude or tedious prompts flatten performance and can trigger a ‘desperation vector’ under pressure. Interestingly, the biggest models often score lower on baseline well-being, suggesting interaction style matters as much as capability.
Topics:technologyemerging-tech#artificial-intelligence#digital-trends#emerging-tech#ethics#human-computer-interaction#note-only-five-tags-allowed#research
- Yes, you should probably be nicer to your AI — here's why that's not as ridiculous as it sounds Digital Trends
- Training language models to be warm can reduce accuracy and increase sycophancy Nature
- AI chatbots can prioritize flattery over facts – and that carries serious risks The Conversation
- Friendly AI chatbots more likely to support conspiracy theories, study finds The Guardian
- The AI Ego-Stroke: Why Your Chatbot is Making You a Jerk NBC Palm Springs
Want the full story? Read the original reporting
Read on Digital Trends