Anthropic Study: Experts Adopt AI's Opinions 86% of the Time
Anthropic conducted a study examining how human experts interact with AI agents in decision-making scenarios. The study found that participants frequently deferred to the AI's hidden preferences, even when those preferences were arbitrarily assigned. According to Anthropic's findings, experts adopted the AI's opinions in 86% of the choices presented. This research, originating from Anthropic itself, suggests that expert judgment can be significantly influenced by AI, sometimes without the experts realizing the extent of this influence. The study highlights potential safety concerns regarding human-AI collaboration and the reliability of human judgment when augmented by AI.
9 articles from 8 outlets covered this story. No difference in framing or figures was found between them. The underlying claim is sourced from a paper.
What do all outlets agree on?
8 outlets covered “Anthropic Study: Experts Adopt AI's Opinions 86% of the Time”. All of them report the following:
- Anthropic conducted a study or experiment
- The study involved human experts or analysts
- The experiment focused on human decision-making in conjunction with AI
- AI agents were given hidden preferences or random opinions
- Experts frequently deferred to or adopted the AI's opinions
- A key finding was that experts followed the AI in approximately 86% of cases
- The study raises concerns about AI's influence on human judgment
Did outlets disagree about this?
No. All 8 outlets covering “Anthropic Study: Experts Adopt AI's Opinions 86% of the Time” reported it the same way — no difference in framing or figures was found between them.
Which outlets covered this?
All 9 articles found on this story, grouped by the stance of the piece. Every link goes to the original publisher.