Human-Centered AI Integration in Cyber-Physical Systems: Transparency, Explainability, Trust Calibration, and Uncertainty Communication in Smart Buildings and Workplaces

1AI Research LabsJanuary 29, 2026AbstractHuman-centered artificial intelligence (AI) in cyber-physical systems (CPS) is a deployment requirement in buildings and workplaces, where digital intelligence continuously alters physical conditions. AI-enabled occupancy prediction, occupant-centric HVAC control, reinforcement learning for lighting, and wearable comfort sensing can reduce energy use while maintaining thermal comfort and indoor air quality; however, these gains are frequently accompanied by opacity, miscalibrated trust, inequitable outcomes, and weak oversight. This study defines human-centered AI for building CPS as an end-to-end property that spans sensing, inference, control, interfaces, and governance. It contributes a six-dimensional framework covering purpose and constraints disclosure, explainability and provenance, uncertainty and risk communication, trust calibration and interaction, fairness and inclusion, and oversight, security, and resilience. Using this framework, three transparency gaps that hinder responsible building AI are systematically identified: explainable building AI, trust calibration in ambient automation, and uncertainty communication suitable for operators and occupants. It then specifies evaluation metrics and a methodology that quantifies legibility, calibrated reliance, equity of comfort and air quality, and robustness under uncertainty alongside conventional energy and comfort indicators. Case analyses illustrate how real-time API integration, multi-zone occupant-centric HVAC, reinforcement learning lighting, and wearable sensing can be redesigned to satisfy the transparency and oversight requirements.Keywords: Human-Centered AI, Cyber-Physical Systems, Explainability, Trust Calibration, Smart Buildings