Artificial Intelligence for Smaller Manufacturers
In simple and unscientific terms, artificial intelligence (AI) is the programming of a machine to "think", work and react as if it were a human. It is another tool for us, but one that acts as another person in our office or factory.
The technology is becoming more accessible to small companies as larger tech companies, like Microsoft, build technology to help smaller entities build their own AI models. These models can do many things including document processing, image classification, text recognition, sentiment analysis, prediction, and text translation just to name a few.
What is the potential?
When contemplating artificial intelligence for industrial applications, one may quickly think of the potential benefits to a current worker. Imagine you have a database of problems and solutions that you have gathered from years of experience in business. Then consider the possibilities if you combine that database with an AI model that processes an image of a problem to point you in the direction of a possible solution. That worker can harness the knowledge and expertise of their aging and retiring colleagues while lightening their own cognitive and decision-making load.
I am personally drawn to the potential of having a seasoned professional in my pocket who knows what a good job is supposed to look like. Today, I'll take a photo of something and send it back to the shop where some of our more experienced woodworkers chime in about what the issue is and what possible solutions might be. But what if there were an AI model that could respond without direct human intervention, or offer inspiration or another point of view than what the employees of one company might be able to figure out or recommend?
When I tour the great cathedrals in Europe, the ones that take hundreds of years to build, I cannot help but think of the knowledge each one of those workers who built it possessed. That knowledge may have been passed down, but in the time since those churches were built to our present day, how much of that knowledge has been lost? I see AI solutions to help us capture the knowledge we have today so future generations can apply it to their own problems. I do not think that the way they built cathedrals in the past is the only way to build a structure, but I do think that having a deep understanding of those means and methods and the ability to tap into that knowledge can be extremely powerful.
A more practical ability for AI today is in the processing and analysis of construction documents. Imagine that you are an estimator. You see many different projects over the course of a month or a year. These projects are drawn by different architects, using different ways of communicating project requirements. It takes you many estimates to recognize patterns and develop a mental model of how to find the required information from a given architect.
Now imagine that you have an AI assistant that can analyze documents for you so you can focus on higher level problems, such as helping to smooth out some of the discrepancies and offering solutions to the clients. If you are spending your time estimating by performing a take-off (counting and logging products) then you are missing out on places you can add value to the construction process. Especially as someone who is looking at the project for the first time in your company, the question should be "how can your company add value to the project and help achieve the client's goals?"
My Experience
I recently discovered Microsoft's AI Builder as part of the Power Platform. I started playing around with the solution by importing invoices. My goal was twofold. I wanted to check the invoice against the items we ordered and I wanted to establish a process that would get done without relying on an individual's energy, discipline and time.
The interface and solution that Microsoft developed makes it very easy to upload and begin analyzing the documents. There are definitely flaws in its ability, but it seems like the more data we upload, the better the output.
The next step would be to connect this model to an automation that will communicate essential information to the team while also reconciling the data with our purchasing function. Stay tuned.
It can be very easy to ramp up one's excitement when contemplating the possibilities surrounding AI. Like a lot of technology, there is a lot of hype and there is an extraordinary amount of work that needs to be done before this type of technology can be a tool that we confidently rely on. But I am planning for the days when the best minds in our shop today are helping the best minds in our shop 200 years from now.