Artificial intelligence dominates today’s headlines, from predictions of “superintelligence” to ad campaigns and daily industry news. But understanding its real-world impact on operations can be challenging.
It can be difficult to figure out where the hype ends and actual day-to-day impact begins. This article explores how AI is intersecting with operational technology (OT), where it is already creating value, and what organizations can do to adopt it effectively.
The big picture? AI isn’t replacing know-how and judgment in OT, but it is augmenting capabilities.
When AI Meets OT
Where it all begins is applying AI techniques and technologies within manufacturing and critical infrastructure environments, explains CoreX Global Head of Innovation Jay Wigard, adding, “The application of AI in OT is one of the big components of ‘Industry 4.0’ (or the ‘Fourth Industrial Revolution’).” And it’s quickly evolving, as industry continues a rapid evolution.
OT environments have several areas of significant importance to manufacturing, including safety, quality, cybersecurity, efficiency, and uptime. To maintain high standards across these critical areas, organizations are currently making big investments to better understand how AI can improve industrial processes and provide innovation.
Illustrating how this looks in action, Wigard provides two examples:
- Quality Control: Computer vision and machine learning identify defects in real time, improving Overall Equipment Effectiveness (OEE).
- Predictive Maintenance: AI analyzes sensor data to forecast maintenance needs, helping to prevent downtime and extend asset lifespan.
AI as a Practical Tool in the OT Toolkit
When thinking about AI, it’s important to understand that it’s another addition to an OT toolbox for solving problems and improving workflows. Are people using AI in OT already? The answer is a resounding “yes.”
To this, Wigard says, “I remember commercials I saw for cleaning products when I was growing up, where the video editing does a funny wipe transition from a dirty, messy house to a clip where that same house is perfect and shiny and clean, with the colors of everything instantly more vibrant.”
He adds, “AI isn’t that! It won’t immediately fix process issues or data issues. There’s still some foundational work most companies need to do to get value out of AI. But it does have the potential to revolutionize and transform workflows and automation.”
Where AI and OT converge is in how strengths and needs really line up. AI is great at sifting through lots of data, identifying trends, and synthesizing or summarizing data. These capabilities are a natural fit in industrial processes with a lot of technical information and data required to efficiently scale.
Wigard again points to predictive maintenance as an area where AI can improve safety and reduce or prevent unplanned downtime, which can cost upwards of $1 million per hour in a manufacturing environment.
(No. We’re not exaggerating that claim.)
AI can accelerate knowledge management and root cause analysis and improve workforce efficiency in operational environments. People retire and leave organizations, and AI becomes that knowledge bridge to help newer workers quickly ramp up on the job and become more efficient faster.
A third area where AI makes an impact is threat detection in industrial cybersecurity. AI can establish a baseline of what is normal in OT networks and systems, quickly detect anomalous activities and attempted access, and even act on those threats.
“I’m really big on looking at how AI can improve the human condition, and there are a lot of practical applications in OT and manufacturing environments,” says Wigard. "OT environments have a lot of complexity to them, which inherently makes the workload stressful. By identifying the areas where AI can help minimize the grind or the risks and the stress, it can improve the human experience of people performing those jobs and managing the processes.”
He adds that CoreX has worked on solutions where AI is used to ingest OEM equipment manuals, knowledge articles, and process best practices and then have ServiceNow’s NowAssist LLM synthesize that material to automatically provide recommended actions to quickly resolve unplanned downtime in the ServiceNow OT incident form.
“We call this the OT Rapid Response use case, and we worked on it with a German auto manufacturer and used technology from ServiceNow and NVIDIA,” explains Wigard. “Even just being able to use ServiceNow as a platform to pull together OT Knowledge from multiple sources is extremely helpful in accelerating OT Incident resolution and helping to quickly enable and train employees.”
Understanding Challenges, Implementing Solutions
The practical challenges of AI in OT are very similar to AI in IT. These include data quality, legacy systems integrations, and cybersecurity risks. The difference is, in OT, these challenges and risks are often magnified. Wigard points out that cybersecurity risks in OT “can be terrifying. What if a bad actor gains access to systems that control water purification or construction equipment?”
The solution to these challenges is a strategic approach to AI implementation. This means assessing AI readiness is a key step in an AI deployment. Identifying blockages in the organization and creating AI remediation plans can help overcome issues like inconsistent data quality or unavailable data.
When rolling out AI, it’s essential to proceed carefully and intentionally with small pilot programs and ongoing governance, observation, and human oversight, says Wigard, adding one key point: Human oversight will be a critical component of AI success, regardless of the industry.
“Small, safe pilots are essential. It’s important to set a strategy and find a platform that can help you execute on the strategy,” he states. “I think ServiceNow is well-positioned to be the platform that helps organizations overcome a lot of challenges, like integration with legacy systems, for example.”
AI isn’t a magic wand in OT or any other business area. But it is a powerful resource that will augment current capabilities and build efficiencies. The value AI can provide in an OT environment isn’t a black box of mystery, and Wigard lays out a roadmap for a successful AI deployment.
“Find quick wins, but don’t get lulled into a false sense of security. Keep in mind you still need people monitoring AI inputs and outputs,” says Wigard. “Stay vigilant, establish effective governance and oversight to make sure AI quality doesn’t drift. Don’t take human oversight out of the equation or assume that AI can replace human intelligence. It should be a ‘better together’ story.”
