Towards User-Centered Metrics for Trustworthy AI in Immersive Cyberspace

Trustworthiness of AI in immersive environments requires urgent attention: The paper emphasizes that AI failures in emerging immersive ecosystems like the metaverse can cause "serious user harm," making trustworthiness metrics essential for these new platforms, not just traditional cyberspace applications.

Existing trustworthiness frameworks are insufficient for immersive technologies: Current approaches to measuring trustworthy AI in fairness, privacy, and robustness were designed for simpler systems and don't adequately address the "much more complicated" performance and user experience assessment challenges posed by immersive ecosystems.

Three core dimensions of trustworthy AI need focus: The paper identifies fairness, privacy, and robustness as critical pillars for evaluating trustworthy AI, suggesting these three areas should be prioritized when developing metrics for immersive platforms.

User-centered metrics are essential, not just technical ones: Rather than relying solely on technical performance measures, the research agenda calls for metrics that directly reflect actual user experiences and needs in immersive environments, recognizing that user impact should drive trustworthiness assessment.

A systematic research agenda is needed for immersive AI governance: The paper calls for developing a coordinated, comprehensive approach to trustworthy AI specifically designed for immersive cyberspace, indicating that ad-hoc solutions won't adequately protect users in these emerging ecosystems.