Building a safer path to autonomous industrial AI
Industrial AI is entering a new phase. After decades of predictive analytics and other specialized applications, advances in foundation models, physical AI, and agentic AI are making it possible to automate more complex tasks across industrial environments.
Industrial AI is entering a new phase driven by advances in foundation models, physical AI, and agentic AI. Foundation models are large AI systems trained on vast amounts of data that can perform a wide range of tasks, while physical AI refers to AI systems designed to interact with real-world machinery and equipment. Agentic AI systems can make decisions and take actions autonomously. Unlike digital-only AI, industrial AI interacts directly with physical systems, such as factory equipment or power grids, where mistakes can have serious safety or economic consequences. According to AVEVA’s chief technologist Arti Garg, the challenge is balancing these new capabilities with safety, reliability, and human oversight. The adoption of industrial AI has accelerated rapidly, with a reported 78% increase over the past two years, reflecting growing confidence in its potential despite its risks.
Industrial environments generate vast amounts of data from sources like telemetry (real-time sensor readings), service logs, and engineering documents. Historically, integrating and analyzing this data has been difficult and time-consuming, often requiring human operators to manually correlate information to diagnose problems. Newer technologies, such as graph databases and AI tools, can now connect and analyze these disparate data sources much faster. For example, an operator could use a tablet with an AI assistant to instantly gather and correlate data from a malfunctioning pump, including its maintenance history and design specifications. AI-powered robots and drones can further enhance this process by entering hazardous areas to collect data without endangering workers. This real-time support improves problem-solving efficiency and safety in industrial settings.
AVEVA has developed a framework for responsible AI that prioritizes security, efficiency, and human safety. This framework includes a multi-layered governance model for both internal AI use and customer-facing AI products. The core principle is that AI should augment rather than replace human workers, particularly in critical decision-making processes. Human judgment and oversight remain central, with guardrails determining when AI can act autonomously and when human supervisors must intervene. Responsible AI also addresses environmental efficiency, as AI systems themselves consume energy and resources. AVEVA is involved in an IEEE working group developing a standard methodology to measure AI’s environmental impact across factors like electricity use, energy consumption, water usage, and carbon emissions.
Industrial AI poses unique risks because it interacts with physical systems that can affect safety, infrastructure, and operations. Unlike digital AI, errors in physical AI can have immediate, real-world consequences, such as equipment failure or environmental damage. Newer AI systems, including agentic AI, are harder to predict and explain because their behavior can evolve over time as they learn. However, these systems also offer significant benefits. For example, they can learn from experience faster than humans and apply that knowledge across multiple sites, scaling expertise efficiently. This is particularly valuable in industries facing workforce challenges, such as the impending retirement of nearly half of the industrial workforce within the next five years. Capturing and applying this expertise through AI could help bridge knowledge gaps and improve productivity.
AI can play a critical role in making industrial processes more sustainable, particularly in managing complex power systems as renewable energy sources like rooftop solar become more common. The electric grid faces challenges in balancing intermittent power generation with demand, maintaining power quality, and detecting anomalies early. AI systems can help grid operators monitor the grid in real time, identify potential issues, and respond faster to disruptions. For instance, AVEVA is collaborating with Idaho National Laboratory on a project to enhance AI-driven grid resilience. By improving the management of renewable energy integration, AI can contribute to more efficient and environmentally friendly industrial operations while ensuring reliable power supply.
The next phase of industrial AI could see greater adoption of autonomous systems, including robots and drones, which are expected to transform operations in factories, power systems, and mining sites. These systems promise increased efficiency, safety, and productivity by reducing the need for human workers to enter hazardous environments. However, realizing this potential requires more than just deploying new technology. Organizations must rethink business processes, establish robust safeguards, and provide workers with new tools to apply their expertise in an AI-enhanced environment. The goal is to create a model of automation that is not only more autonomous but also safer, more efficient, and more sustainable. As Arti Garg notes, these changes will fundamentally alter how work is done in industrial settings.
