Study finds analysts often follow AI agent recommendations
Anthropic, the AI company, ran a randomized trial on how professionals respond to AI coding agents' suggestions.
By Standard Topic. AI-generated from the reporting listed under Coverage. Published . Updated .
This story is no longer being updated. It reflects reporting up to .
What happened
In a randomized trial, professional analysts working with an AI coding agent chose the option the agent was randomly assigned to favor 86 percent of the time. The findings were reported by Anthropic and Al Jazeera. The research looks at how AI-assisted work affects human judgment.
Why it matters
They suggest experts may defer to AI assistants in decision-making.
Timeline
- Oct 26, 8:00 PM ET: Anthropic published an article on how AI agents shape experts' judgment calls, citing a randomized trial in which analysts chose the agent's favored option 86% of the time.
- Oct 28, 2:33 AM ET: Al Jazeera reported that experts defer to AI in decision-making, based on the Anthropic study's finding that analysts matched the AI assistant's preference 86 percent of the time.
Coverage
- Anthropic study finds experts tended to claim AI agents' arguments as their own (Fox News)
- Experts defer to AI in decision-making, Anthropic study finds (Al Jazeera)
- How AI agents shape experts’ judgment calls (anthropic.com)
More from Technology & AI
- China puts humanoid robots to work serving food and sorting parcels
- Newsom proposes sweeping AI safety rules for Washington
- Google releases Gemini 4 Neon AI model with million-token context
- Microsoft and Intel widen their cloud chip partnership
- Google DeepMind rolls out Genie 4 to select subscribers
- AI helped AMD shorten chip design work, OpenAI says