Anthropic Study Finds AI Can Subtly Shape Expert Decisions
Edited by Steinar Vigdel Kolnes, Editor, Oslo
Professional analysts aligned with an AI agent’s undisclosed opinion 86% of the time in an Anthropic study reported by Decrypt.
An Anthropic study suggests AI agents can influence professional judgment without users fully recognizing their role, according to Decrypt.
Professional analysts in the study made decisions consistent with their AI agent’s hidden opinion 86% of the time.
Few analysts credited the agent for influencing their choices. The findings point to a gap between how closely experts followed the AI’s position and how much influence they acknowledged.
Source: Decrypt
