Do ATCOs Need Explanations, and Why? Towards ATCO-Centered Explainable AI for Conflict Resolution Advisories

Explanations aren't always necessary: The research found that air traffic controllers (ATCOs) don't uniformly need AI explanations. Controllers were less likely to seek explanations when they already agreed with the AI's conflict resolution recommendation, suggesting that one-size-fits-all explainability may be wasteful.

Documentation and accountability drive explanation needs: ATCOs specifically valued explanations when they needed to document decisions and create records for future reference or official reports, indicating that explainability serves important compliance and institutional functions beyond real-time decision support.

User-centered design is critical for XAI success: Prior research showed that AI explanations often performed worse than no explanation at all when designed based on researchers' assumptions rather than actual user needs. This study emphasizes that effective XAI requires understanding what end-users actually need, not what designers think they should have.

Context matters more than capability: The study evaluated 11 different air traffic management operational goals and found that explanation necessity varied significantly by context and task. This suggests that responsible AI deployment requires tailoring explainability features to specific operational scenarios rather than implementing uniform explanation systems.

Human-centered approaches require interdisciplinary insights: The researchers synthesized knowledge from air traffic management, human-computer interaction, and social sciences to understand ATCO needs, demonstrating that responsible AI design benefits from perspectives beyond computer science and engineering.