The Persuasion Paradox: When LLM Explanations Fail to Improve Human-AI Team Performance

Research reveals a 'Persuasion Paradox' where LLM explanations increase user confidence but don't reliably improve human-AI team performance, and can actually undermine task accuracy. The study found that explanation effectiveness varies significantly by task type, with visual reasoning tasks seeing decreased error recovery while logical reasoning tasks benefited from explanations.

• →LLM explanations systematically increase user confidence and reliance on AI without consistently improving task accuracy.
• →For visual reasoning tasks, explanations suppress users' ability to recover from AI model errors compared to probability-based interfaces.
• →Language-based logical reasoning tasks showed improved accuracy with LLM explanations compared to other support methods.
• →Subjective metrics like trust and confidence are poor predictors of actual human-AI team performance.
• →Task-dependent and cognitive modality factors strongly influence the effectiveness of AI explanations.