AI in Engineering: The Challenge is No Longer Possibility, but Deployment

Artificial intelligence has moved beyond the point where engineering businesses need to be persuaded that it matters. The more difficult question is what to do with it.

Manufacturers and engineering teams are surrounded by possible applications: machine vision, predictive maintenance, process optimisation, design support, quality, planning, controls, documentation and compliance. That breadth is exciting, but it can also make adoption harder. When almost anything can be described as an AI opportunity, choosing where to begin becomes a management challenge in its own right.

Jon Excell, Editor of The Engineer, believes there remains a significant gap in practical understanding. He points not only to uncertainty about what AI actually is and where the opportunities lie, but also to the practicalities of deployment, ethics, and skills. That is an important distinction. An impressive demonstration is not the same as a solid industrial application.

Isabelle Russell, UK Aerospace Account Manager at Siemens, argues for a more incremental route. Digitalisation, she says, does not have to be “all or nothing”. Small interventions can remove individual roadblocks while gradually building the digital architecture on which more advanced capabilities can sit.

That is particularly relevant to established engineering businesses. Most are not starting with a clean sheet, and many operate long-lived assets, fragmented software systems and highly regulated processes. Successful AI adoption, therefore, depends on the quality and accessibility of underlying data, the ability to connect systems and a clear understanding of the decision or process being improved.

The most valuable application may not always be the most obvious one. Stephen Phipson, CEO of Make UK, highlights the potential of AI in compliance, particularly in regulated sectors such as aerospace and medical devices. Productivity on the shop floor matters, but regulated businesses also carry a substantial burden of evidence, documentation and verification. If AI can support those activities safely and reliably, the commercial impact could extend well beyond incremental cycle-time improvements.

This does not mean handing regulatory judgment to an algorithm. Engineering organisations still need traceability, validation and accountable human decision-making. The opportunity lies in using AI to interrogate information, identify inconsistencies, support evidence gathering, and reduce manual effort in navigating complex requirements.

Elsewhere, AI is increasingly intersecting with traditional control engineering. Abhijith Sreekumar of Lotus Engineering sees opportunity in bridging machine learning and control systems, while Jonathan Cooper of Williams Grand Prix Technologies points to physical machine learning as a technology with potential across a wide range of engineering problems. These are not separate digital experiments. They influence how products behave and how engineering decisions are made.

Automation is evolving in parallel. Dave Sutton of Schneider Electric argues that industrial automation remains constrained by proprietary legacy technology and sees software-defined automation as a path to greater openness. That matters because AI is far more useful when information can move across the operation. A model that identifies a problem but cannot connect effectively with the production environment has limited value.

For many organisations, the next sensible step will not be a company-wide AI programme. It will be one well-chosen use case with a clear problem, good data, measurable outcomes and an implementation team that understands both the technology and the engineering context. Prove that, learn from it, and the route to wider adoption becomes considerably clearer.

The AI, Digitalisation and Automation track at Advanced Engineering 2026 will focus on this gap between possibility and feasible deployment. The important conversations are not about whether AI will influence engineering. They are about where it produces measurable value, what data and architecture are required, how established operations can adopt it without needless disruption, and where human engineering judgement remains essential.

Advanced Engineering takes place at the NEC Birmingham on 4-5 November 2026.