Public Utilities

Review Examines Human-Governed Agentic AI for Water-Utility SCADA

A review of agentic AI for water-utility SCADA systems recommends gradual, human-governed deployment with monitoring, explainability, and rollback controls.

A scoping review examines human-governed agentic AI for supervisory control and data acquisition systems in water utilities. It evaluates supervised, semi-supervised, unsupervised, graph-based, digital-twin, and reinforcement-learning approaches under different data and deployment conditions. The review concludes that these methods can support anomaly detection and decision-making but should not replace operator judgment in water-process control. It recommends gradual deployment, beginning with rule-based checks and thresholds before introducing machine-learning systems and agents as data quality improves. The study also identifies operational safeguards including false-alarm monitoring, drift detection, explainability, rollback procedures, and periodic operator review. The development matters because it provides a structured framework for adopting AI in utility control rooms while limiting the risks of unsupervised automation.

Original source: Springer Nature

Related RAQI solutions

In-depth analysis on RAQI Insight

Work with RAQI