Make UK has warned that manufacturers risk losing productivity gains from artificial intelligence unless factory-level skills, training capacity, and adoption support improve.
The organisation’s latest report says AI is still concentrated in back-office functions rather than core production environments. Among manufacturers already using AI, 83% are applying it in areas such as HR, finance, and administration, while operational adoption remains much lower. Only 11% are using AI in production, 7% in supply chain and logistics, and 6% in quality control.
Skills shortages sit at the centre of the report. More than half of manufacturers identify capability gaps as their main barrier to using AI more effectively, with the pressure particularly visible at technician and operator level. The report says manufacturers are not simply looking for advanced coding capability, but for practical skills in data literacy, problem solving, leadership, and change management.
Those capabilities determine whether AI tools can be understood, challenged, and embedded in production routines. In manufacturing, much of the value is expected to come from better use of production data, predictive maintenance, scheduling support, anomaly detection, visual inspection, process optimisation, and decision support. Each use case depends on people who understand the process well enough to interpret outputs and spot weak assumptions.
Make UK estimates that the manufacturing sector loses around £6bn in output each year through unfilled vacancies and digital capability gaps. It also argues that wider digitalisation across the sector could add £150bn to UK GDP by 2035, provided adoption moves beyond isolated pilots and administrative tools.
The recommendations include nationally recognised AI skills standards for manufacturing roles, more practical support for SMEs, flexible training that can work around shifts, and stronger use of programmes such as Made Smarter. Make UK also wants the Advanced Manufacturing AI Champion to help translate national AI policy into support that can be used inside factories, rather than leaving smaller companies to interpret general technology ambition on their own.
Manufacturers are under pressure to increase output, reduce waste, improve resilience, and absorb higher operating costs while dealing with persistent workforce shortages. AI is frequently presented as a route to productivity, but factories rarely operate in the clean conditions assumed by early pilots. Production sites contain legacy machines, uneven data quality, site-specific processes, supplier variation, and safety-critical procedures that cannot be redesigned at speed.
The skills gap is therefore an operating constraint rather than a separate workforce issue. Maintenance teams using predictive analytics still need equipment knowledge. Planners using AI-assisted scheduling still need judgement about labour, materials, customer commitments, and disruption. Quality teams moving towards exception-based inspection still need to understand process drift, tolerance risk, and failure modes.
Tacton’s 2026 manufacturing study points to a similar weakness across product lifecycle systems, with manufacturers investing in AI while many still struggle with inconsistent configuration logic across sales, engineering, and production. AI can accelerate useful work when the underlying data and workflows are coherent, but fragmented systems create a route to faster errors.
Training is also harder to deliver in factories than in office environments. Time away from the line competes with production schedules, overtime, maintenance windows, and customer deadlines. Smaller manufacturers face an additional burden because operational, commercial, quality, and improvement responsibilities are often carried by the same small group of people.
AI adoption in manufacturing is now moving into a more practical phase. Early curiosity around generative tools is giving way to tougher questions about return on investment, governance, cyber security, traceability, safety, and integration with operational technology. Companies likely to benefit first will be those that link AI to measurable constraints such as scrap reduction, lower downtime, faster quoting, improved inspection throughput, or more reliable scheduling.
For the UK, national AI ambition will have limited industrial value unless manufacturers can train people close to the process, identify use cases that survive production reality, and access support that recognises SME operating pressures. The technology is advancing quickly, but the bottleneck is increasingly human, organisational, and practical.
