How Industrial AI Can Scale Without Compromising Safety
Industrial AI is moving toward more autonomous tasks, but deploying it safely requires better data integration, governance, and human oversight, according to AVEVA's chief technologist. Because industrial AI can directly affect physical systems, unpredictable decisions can threaten safety, reliability, and critical infrastructure. The article also covers using AI to assist operators, collect data in hazardous settings, and measure its environmental impact.
AVEVA chief technologist Arti Garg tells MIT Technology Review Insights that industrial AI is moving beyond predictive analytics toward foundation models, physical AI, and agentic systems. Because these tools can act on physical equipment, unexpected decisions may affect safety, reliability, and critical infrastructure.
The discussion points to combining sensor readings, maintenance records, and technical documentation; using robots to inspect hazardous sites; and governance that keeps people involved in consequential choices. It also notes an IEEE group developing a way to measure AI’s power use, energy, resources, water, and carbon impacts.
If industrial AI becomes more autonomous, plant operators, utility workers, miners, and nearby communities could feel the effects. Safer robots and better data integration may reduce exposure to hazardous tasks, while stronger governance and human oversight could help prevent physical failures. However, the benefits may depend on workforce training, clear accountability, and environmental measurement. Workers might see roles change rather than disappear, with domain expertise becoming more important for supervising and guiding automated systems.