People Attribute Purpose to Autonomous Vehicles When Explaining Their Behavior: Insights from Cognitive Science for Explainable AI

People naturally explain autonomous vehicle behavior by attributing purpose and goals to them. The study found that participants predominantly used "teleological" explanations—attributing intentions and purposes to the vehicle (e.g., "the car is trying to avoid the pedestrian")—rather than purely mechanical descriptions of how the system works.

Purpose-based explanations are perceived as higher quality by the public. When researchers tested how people rated different types of explanations, teleological explanations were rated significantly better than counterfactual ones (which describe what would have happened under different conditions), suggesting that people find purpose-driven narratives more satisfying and trustworthy.

Designing AI explanations based on how humans naturally think may improve public trust. The paper argues that effective explainable AI should align with human cognitive patterns rather than purely technical descriptions, implying that XAI systems for autonomous vehicles should incorporate purpose-based language to better communicate with users.

Current XAI design may not match human reasoning patterns. The research reveals a gap between how AI systems are typically explained (mechanistically) and how people naturally understand and prefer to understand them (teleologically), highlighting a potential mismatch in current explanation approaches.

Empirical testing of explanation styles is essential for responsible AI deployment. By conducting controlled experiments with real participants, the researchers demonstrated that responsible AI requires grounding design choices in evidence about human cognition rather than assumptions about what explanations should work.