Somewhere right now, a company is explaining how artificial intelligence will transform the offsite construction industry. It will help schedule production, track materials, prepare estimates, answer customer questions, and perhaps eventually explain why the forklift is never where anyone left it.
Meanwhile, a new employee on a factory floor is trying to figure out which drawing to use. The one on the table looks different from the one attached to the wall, and the person who knows the answer is at lunch. He spent yesterday following someone around and was told this morning he should be getting the hang of it.
I’m interested in what AI can do for our industry. There are useful possibilities, and some deserve serious attention. But I keep coming back to a fairly ordinary question: Why do we spend so much time discussing how to make machines smarter and so little time discussing how to help people get better at their jobs?
AI Has a Marketing Department. Training Has Joe.
Part of the explanation is straightforward. AI companies have products to sell, so they produce demonstrations, sponsor events, publish articles, and explain what their technology might accomplish. There is nothing wrong with that. Selling things is how businesses stay in business.
Human training, on the other hand, often has Joe. Joe has worked at the factory for twenty years, knows where everything belongs, and can recognize a problem before most people realize there is something to look at. Naturally, when someone new arrives, management says, “Put him with Joe.”
Unfortunately, nobody asks whether Joe knows how to teach. Nobody adjusts his production expectations, either. He is supposed to complete his work, answer questions, explain procedures, prevent mistakes, and somehow turn the new person into a productive employee by Thursday.
Joe may be a wonderful craftsman. That doesn't automatically make him an instructor, just as owning a piano doesn't make someone qualified to give lessons. And when his explanation begins with “You just do it like this,” the new employee may need a little more information.
Purchasing Is Easier Than Teaching
Buying technology comes with a reassuring sequence of events. There is a presentation, a proposal, an agreement, and usually a meeting where somebody says “implementation.” Management can point to a decision and a budget and explain that the company is moving forward.
Training people is less tidy. It requires someone to explain the work, demonstrate it, watch another person attempt it, correct mistakes, and come back later to see what stuck. Sometimes the explanation needs to change because the first one made perfect sense to the instructor and absolutely none to the employee.
It also takes time when everyone already claims to have none. A factory running behind schedule can struggle to set aside thirty minutes for instruction, even while spending hours correcting work that better instruction might have prevented.
That is an expensive little circle to keep walking around.
Before Teaching the Standard, Agree on One
Another reason training can be uncomfortable is that it reveals how much management has left unsettled. The moment someone tries to write down the correct way to perform a task, questions start to appear.
Which drawing controls? Who approves a substitution? What does an acceptable finished connection look like? When should an employee stop working and call a supervisor? Why does the morning supervisor want something done differently from the afternoon supervisor?
Those are management decisions. A new hire cannot settle them by paying closer attention.
Picture an employee being told to follow the established procedure, only to discover that the procedure changes depending on who walks past the station. Before long, that person learns to watch for supervisors instead of watching for quality. That is certainly a learned skill, although probably not the one the company intended to develop.
Creating a useful training program requires management to agree on what people should learn. That's often the hardest part of the entire exercise.
Attendance Is Not the Same as Ability
We also need to be careful about what we call completed training. Gathering employees in a room, showing slides, and passing around a sign-in sheet proves that a meeting took place. It does not necessarily prove that anyone can perform the task afterward.
The useful question is whether an employee can demonstrate the work, explain what could go wrong, and recognize when help is needed. That takes observation, practice, and a supervisor willing to ask something more revealing than “Everybody understand?”
Most people know the answer expected when that question comes at the end of a meeting. They also know lunch starts in four minutes.
A better approach is to ask someone to show the process and explain it in their own words. Any confusion becomes an opportunity to improve the instruction before it becomes a problem inside a finished wall.
Training Costs Show Up Immediately. Mistakes Arrive Later.
The cost of teaching someone is easy to see. Two employees spend time working through a task, and production may be lower during that period. Put it on a spreadsheet, and it looks like an expense.
The cost of inadequate training is harder to gather in one place. Some of it appears as wasted material. Some appears as rework, overtime, inspection problems, or service calls. Some is hidden in a supervisor answering the same question for the seventh time that week.
Without connecting those costs, an owner can conclude that training is too expensive while still paying for its absence.
Factories don't need an elaborate corporate academy to start improving this. Pick a task that repeatedly causes trouble. Establish the correct method, give someone time to teach it, let employees practice, and check whether the results improve. A clear instruction sheet at the workstation can be more useful than a thick training binder quietly collecting sawdust in the office.
Give AI a Job in the Training Process
This is actually an area where AI could earn its place. It can turn approved procedures into readable instructions, prepare practice questions, or draft explanations for employees at different experience levels. It can also help prepare translations, provided someone qualified checks the meaning, especially where safety and technical details are involved.
The important word is “help.” Someone who understands the work still needs to verify the material and determine whether the employee can apply it correctly.
An AI-generated instruction can sound wonderfully confident while missing a detail that matters on your particular production line. A polished explanation isn't necessarily correct just because it has numbered steps and excellent grammar.
Used thoughtfully, these tools could give experienced employees more time to demonstrate, coach, and answer questions. That would be a worthwhile improvement for both Joe and the person following him around.
Gary’s Observation
I would like to hear more factory owners ask, “What could our people do better if we taught them properly?” That question belongs in the same conversation as software, automation, and the next equipment purchase. It deserves a budget, an owner, and time on the schedule.
Employees need clear expectations, useful instruction, and the opportunity to ask questions without feeling foolish. Supervisors need help becoming better teachers, and experienced workers need time to share what they know before that knowledge leaves through the retirement door.
Keep exploring AI. There is plenty worth learning. But before the next presentation about an intelligent factory, walk the line and ask your newest employee how the training is going.
You might learn something that never appeared in the demonstration—and Joe might finally get some help.
No comments:
Post a Comment