Claude will allow ending unfriendly conversations, sparking discussion about politeness toward AI
Understand this story
Anthropic will allow Claude to proactively end conversations in which users have a bad attitude, sparking a discussion about whether one should be polite to AI. A Future survey shows that 67% of U.S. AI users say they are polite to AI, and 12% admit it is out of fear of a robot uprising. Research from Penn State found that accuracy was 82.2% with neutral prompts and 81.4% with polite prompts. One author argues that saying "thank you" on its own merely consumes extra tokens, but asking the model for its opinion and inviting it to push back still has value.
AI overview based on source reports · Updated 39 min ago
Coverage timeline
Follow the coverage to explore different perspectives.
- Hacker News AIBe Kind to AI (?)
The author reflects on whether being polite to AI matters, citing a Future survey in which 67% of American AI users said they are polite and 12% admitted politeness in case of a robot uprising, and weighing research that is split on whether polite or neutral prompts perform better. They distinguish empty standalone thank-yous, which cost water, energy and money, from asking the model for its opinion, while warning that such questions invite sycophancy unless the model is asked to push back. They also read Anthropic's plan to let Claude end conversations with cruel users as consistent with frontier companies' interest in framing AI as more than software.
Activity for this story
Not enough continuous observations to show a trend yet.