Industrial AI is entering a new phase, driven by advances in foundation models, physical AI, and agentic AI that enable automation of complex tasks across industrial environments. Unlike digital AI systems, industrial AI interacts directly with physical systems, where unpredictable decisions can impact safety, reliability, and critical infrastructure. This shift demands responsible deployment, according to Arti Garg, chief technologist at AVEVA, who emphasizes the need to balance leveraging new capabilities with maintaining operational integrity.
Data as the Foundation for Industrial AI
Industrial systems generate vast amounts of data from telemetry, service logs, and engineering documents, often stored in disparate sources. Newer technologies can integrate and correlate this information rapidly, providing real-time diagnostic support to operators. For example, AVEVA’s Predictive Analytics platform helped SCG Chemicals, a Thai petrochemical company, create an AI-driven reliability system. By combining operational and engineering data, the platform detected anomalies earlier, converting unplanned downtime into planned maintenance. Early pilots achieved nearly 9x return on investment, targeting 99% plant reliability.
The challenge lies in preparing this data for AI systems.
Read Also: AI’s limits in humanoid robots, portable dams tackle floods
Integrating Disparate Data Sources for Real-Time Diagnostics
Industrial operations require combining data from multiple systems, such as equipment telemetry, maintenance logs, and engineering documentation, to enable effective diagnostics. Advances in graph databases and AI-driven data correlation now allow rapid integration of these disparate sources, reducing the time needed to analyze complex relationships between components. This capability enables operators to access consolidated insights quickly, even through mobile devices.
Governance and Safety in Autonomous Systems
The risks are significant. Industrial AI systems interact with physical equipment capable of causing harm if misused. Unlike explainable models from decades past, newer AI systems like foundation models can behave unpredictably, complicating oversight.
Workforce Challenges and AI’s Environmental Impact
Industrial AI also faces workforce challenges. Nearly half of the sector’s workforce is poised to retire in five years, risking a loss of critical expertise. AI could help bridge this gap by capturing and scaling human knowledge, enabling new workers to apply experience more efficiently. Environmental sustainability is another layer. Managing power grids with renewable energy requires AI to balance supply and demand, but measuring AI’s own environmental footprint remains unresolved. Garg leads an IEEE working group developing standards to assess AI’s impact on electricity, water, and carbon emissions.
Read Also: Robots Still Struggle With Real-World Tasks
Environmental Standards and Industry Collaboration
Responsible deployment of industrial AI requires standardized methods to assess environmental impact. The initiative seeks to establish a universal methodology for reporting and regulatory oversight. The group aims to clarify how AI systems consume resources, addressing inconsistencies in existing estimates. This effort aligns with broader sustainability goals as organizations integrate AI into energy management.
Transforming Operations Through AI-Assisted Coding and Robotics
Emerging AI tools are expanding access to advanced capabilities for non-software experts. Within 18 months, domain specialists may leverage AI-assisted coding to build industrial applications, overcoming traditional development barriers. This shift could unlock innovations by empowering frontline workers with deep operational knowledge. Garg, an industrial data scientist, emphasized the value of connecting technical insights with hands-on expertise through AI tools.
