A recent Anthropic study involving 50 professional analysts using Claude Code found that experts may defer underdetermined decisions to an AI agent without realizing it.