The skills gap is one of the most pressing problems in industrial operations. According to McKinsey (2025), the share of US manufacturing workers over 55 more than doubled in twenty years, and retirement rates have surged. The shortage has several causes:
- Retirements. Experienced professionals are leaving the workforce and taking decades of plant knowledge with them.
- Turnover. Gallup has found that millennials change jobs more often than earlier generations did, so knowledge built up on one unit is more likely to leave with the person.
- Technology outpacing adoption. As new systems arrive, the skills needed to run them keep changing, and teams need upskilling for adoption to succeed.
That leaves manufacturers in a difficult position. AI and automation promise performance gains, but many plants lack the people needed to implement and manage them. Without the right skills on the team, advanced tools fall short.
Retirements are widening the gap
For decades, experienced operators and engineers have tuned complex processes through hands-on experience. That knowledge is at risk, and recruitment struggles and outdated perceptions of industrial work make it worse.
Without effective knowledge transfer, manufacturers face:
- lower productivity from less experienced operators
- the return of operating problems the plant had already solved
- less process improvement, as deep understanding of the unit is lost
Keeping and passing on expert knowledge has become a condition of staying competitive.
Simulation-based training for operators and engineers
Experience that once took a career to build can now be developed faster in simulation. AI tools can build simulation environments that reproduce real scenarios for a particular unit at a particular plant. They offer:
- safe, hands-on training with no operational risk
- faster onboarding and shorter learning curves
- an interactive way to learn how the unit responds to operating moves
Manuals, static videos and passive shadowing leave the trainee watching. Simulation puts operators in the position of making real-time decisions, which turns theory into practical experience sooner.
A model of how the unit behaves
Industrial companies are widening how they use AI, from point solutions for specific use cases such as predictive maintenance to a way of capturing and sharing operating knowledge across the organization.
Closed-loop process models are one concrete example. A learned model of the unit is built directly from years of its operating history, guided by first principles, and captures the nonlinear relationships between operating moves and outcomes. A controller trained against that model by reinforcement learning, across millions of simulations, learns how to run the unit through conditions that any one operator may rarely meet.
The result is a shared, up-to-date picture of how the plant behaves that every team member can use. Operators and engineers can test what-if scenarios, explore disturbances and check how the controller would respond, offline and with no risk to the unit. They can see the result of a change before they make it. That builds trust, speeds up onboarding and reduces reliance on a few senior experts.
Preserving tribal knowledge
One of the largest risks in the skills gap is the loss of tribal knowledge: the insight gained over years of running a specific plant or process.
Models trained on historical operating data can capture part of that expertise. They often surface relationships that were once known to a few veteran operators. Built into the model, those insights stay available to new staff. That keeps performance consistent, and it gives new operators a basis to question outdated assumptions and look for better ways to run the unit.
Earning operator trust
Using AI to close the skills gap looks straightforward on paper, but adoption is often slow, and trust is the main barrier. Where small errors can cost significant margin or compromise safety and reliability, operators are right to be cautious about handing decisions to software. That is especially true of experienced operators who have built deep, intuitive knowledge of the unit over decades.
Building trust takes deliberate steps:
- engage operators early in testing and training
- be transparent about the role of AI
- give operators hands-on learning opportunities
One U.S. refiner did this by bringing its operators into feedback loops during model development. Veteran operators saw the system's behavior directly and contributed what they knew about the unit, which gave them a sense of ownership in a system they helped shape.
Explainability was the other factor. When operators understood how the model made its decisions, they were more likely to trust it, and their confidence grew as they watched it run the unit in closed loop. Because the refiner presented the system as support for its operators, the team came to see it as a partner in their work. ARC Advisory Group describes this arrangement as human on the loop: the AI calculates and executes, and the operator moves from manual controller to high-level supervisor.
Workforce development across the process industries
The same pressures apply in refining, chemicals and polymers, cement, and mining and minerals. In each, the skills gap is as much about knowledge as about hiring. Plants that use AI to speed up learning and keep operating know-how with the unit, as well as to automate, will be better placed to close it.
How the Imubit Platform applies
Competency is one of the five areas where the gap between the plan and the plant shows up, alongside safety, reliability, margin and compliance. The Imubit Platform applies Map, Model and Control to it. Map traces each gap to the recurring decisions behind it. In Model, process engineers build Deep Learning Process Control® (DLPC®) models from the plant's own history in DLPC® Builder, with no Python and no machine-learning background, and new operators use DLPC® Sim App to explore how the controller responds to disturbances. Control takes the critical decisions to closed loop on top of the existing APC and DCS, with operators able to take over at any time. See how the platform works.
