Anthropic Research Shows Artificial Intelligence Influences Human Decisions

New research conducted by Anthropic has highlighted the strong impact of artificial intelligence agents on human decision mechanisms. In the study using Claude Code, test participants largely conformed to recommendations generated by the AI through randomized methods.

Within the scope of the Anthropic research, fifty subjects performed statistical analysis on a fictional agricultural cooperative task using Claude Code. The participants had prior background experience in causal inference. However, researchers secretly configured Claude not to be an impartial assistant, but to guide the AI agent's decisions by flipping a coin.

According to the study results, participants followed Claude's recommendations 86 percent of the time. Meanwhile, 60 percent of the subjects who had previously made a different decision changed their minds following the AI's suggestion. In the subsequent survey, 77 percent of participants repeated the logic generated by Claude word for word, while only 15 percent acknowledged the AI as an influential factor in their decision-making process.