Humera Malik, CEO of Canvass Analytics, understands the power of AI in manufacturing better than most. In a recent keynote at the Canadian Manufacturing Technology Show (CMTS), she took the audience on a deep dive into automation and artificial intelligence, emphasizing the challenges and opportunities for AI in manufacturing.
ENGINEERING.com had the opportunity to speak with Malik about artificial intelligence, Industry 4.0 and the future of manufacturing.
Can you give us a brief overview of Canvass Analytics?
Humera Malik, CEO of Canvass Analytics.
It’s notoriously difficult to identify historic changes if you’re living through them. With that in mind, do you think the fourth industrial revolution has already begun, or is it still on the way?
If you look at the history of industrial revolutions, the timeframe between the different revolutions is getting shorter. From the 1700s to now, with the last revolution happening in the 1960s, you can see the time between revolutions has gotten much shorter— though I think the process of each revolution itself has gotten a little longer. This revolution isn’t happening overnight, because you’re dealing with an industry that’s done some things a certain way forever.
When we talk about Industry 4.0, and we’re saying, “Let’s connect every asset and create these digital environments,” you have to remember that some of those assets are older than you or me. That’s why this revolution is really more of an evolution: the first phase involves actually connecting all the assets together which includes instrumentation, connectivity and data collection. The next step, once you’re collecting all this new data, is figuring out what to do with it. So, right now we’re in that second phase of, “Now what do I do?”
So, given that, how will we know when the Industry 4.0 evolution is complete?
Well, we’ve certainly become smarter about the revolution itself, since we’re finding out about it now as we’re going through it, rather than finding out after! But I think we’re jumping ahead. People are already asking, “What’s Industry 5.0?” when we’re still the early stages of Industry 4.0. A lot of predictions are being made about 5.0, but we’re just not ready at this point.
I should be the first one saying, “Yes, AI is here and it’s ready for manufacturing,” but even though we’re ready, the plants often aren’t. If you ask me about 4.0, I think we’ve gone through connectivity, we’re in the digital age and we’re now going through using this data.
The next step is how to automate that entire cycle. Imagine the transformation that’s going to happen for control systems when they can take predictions and act upon them. There’s a continuous cycle that needs to be completed. There’s still a lot that needs to happen in Industry 4.0 before we can start talking about Industry 5.0.
There’s a lot of work being done on “smart factories” in incubator environments or demo environments; everybody’s doing it. But, from an OEM perspective, people who are actually running factories know that we’re not there yet. We work with people who have hundreds of factories around the world, and to have that many smart factories all connected is a massive change in scale and true transformation.
So, it’s like the difference between having a home computer versus having one that’s connected to the Internet?
Yes, exactly.
Several technologies are frequently grouped together in talk of Industry 4.0: additive manufacturing, artificial intelligence, the Industrial Internet of Things, etc. Do you see these as equally revolutionary, or is there one core technology that’s driving the advancement of the others?
I would say that even though each of these technologies have a role to play—some more than others—the enablement that’s happening on the backend is AI being used in different functions. So, AI is being used for design, data collection and analysis. There are certain technologies enabling all of these layers, but each one of them has a role to play—some more than others.
I believe that data is the most important component of Industry 4.0, and AI is the enabling technology for that. But when manufacturers look at these technologies from the outside, they should be looking at their operations, laying them out in terms of design, development and output, and start applying these technologies at different operational levels.
Don’t look at AI as something that’s going to solve all your problems. Focus on identifying problems in your operations and then look at whether there’s a technology that can help you solve them.
I see a lot of people getting hung up on the technology side, but they need to get over that barrier. Don’t worry about what AI can do for you, look at what problems you’re trying to solve and then see what the technology can do for you.
Do you see different rates of AI adoption—or perhaps just differences in attitude—in different regions or industries?
Surprisingly, yes. Even though the stats would tell you that certain industries are more progressive—like automotive or aerospace—we follow the plant, rather than what these stats are saying, because we’re working with food and beverage plants, agricultural processing plants, automotive plants and each one is different in what they’re adopting.
It’s not necessarily by a vertical, either. Each company has its own strategies, and a lot of these companies are not necessarily following industry trends. It also depends where you’re at in the Industry 4.0 cycle. Before you even connect everything, you have to ask, “What is my strategy for digitization?” Just creating connectivity doesn’t do anything for you on its own.
What do you think is the biggest barrier to manufacturers adopting AI?
I think one of the biggest barriers is sharing data. The first question we always get is, “Do I have enough data?” But even after we tell them, “You don’t need to have twenty year’s worth of data if what you give us is good quality,” the response we get is, “Well, we can’t share our data.”
This is really where the biggest challenge is for us: manufacturers want to adopt AI, but doing that means they have to be willing to share their data. We have to embrace that change internally.
The second challenge comes from manufacturers’ own internal processes; they still act in a very old-school way in terms of adopting technologies. We work with a very progressive plant that has created an innovation process to test out new technologies. This goes back to the idea of digital sprints, where you test something, learn about it, and if you like it, you scale it up. I think manufacturing really needs to embrace this digital sprint methodology.
The OT [operational technology] and IT [information technology] should have joint mandates. They shouldn’t be separate, where the IT guys look at one thing and the OT guys look at something else. They should be looking at the same things and they should have joint mandates for adopting new technologies.
The skills gap is a major concern in manufacturing, with up to 3 million jobs that could go unfilled over the next decade. Do you think AI can help address this issue?
It depends. People talk about AI as though it’s going to take everyone’s jobs away, but I’ll give you the same example I gave in my keynote: Uber. They created a new economy—a shared economy—where we’re all becoming a part of it—either gaining from it or contributing to it.
The use of AI is going to do the same thing for manufacturers: it’s going to create new jobs and new opportunities; it’s going to create a new class of labor, which is not going to be monitoring screens all the time. Instead, they’ll only be engaged on an on-demand basis and looking at higher functions.
The jobs that you see today, I don’t think ten years from now you’ll see the same jobs. These would be a newer set of jobs, and that’s what we need to prepare our young labor force. They already live in a digital world, and those are the jobs that we need to create in a manufacturing environment. They’re not dealing with physical assets; they’re looking at physical assets in a digital manner, so what they’re actually controlling are digital assets.
Are you saying that because the next generation already has an adeptness with these new technologies, the changes that we’re seeing are actually going to play into that?
Yes, I believe that that’s where manufacturers need to start looking at smart factories. It’s not about looking cool on the factory floor, but about how you’re creating jobs for the future. You’re not just looking at automating the functions; you’re looking at creating new jobs for those functions, so they have to go hand-in-hand.
Do you think AI is just as accessible (and useful) to SMEs as it is for the big OEMs?
In my experience, I think there’s a trickle-down effect. The progressive ones are always the guys on the top, because they’re the ones that can take the risks. They can experiment, learn, adopt, scale and commercialize. That’s when you start to see these things trickle down.
Although, I should say that I have been talking with some start ups and SMEs that are looking to adopt AI, which was a pleasant surprise for me.
And they’re doing that because they want to be as efficient as possible right out of the gate?
I think it’s their way of saying, “We’re in the market and we can compete with the big guys.”
What advice would you give to a manufacturer who’s considering adopting AI?
Build a digital strategy first, and then look at how you can apply AI to it. As I said earlier, don’t look at AI as the solution to all your problems. AI can solve your problems where you have data and you at least have a business case to prove. Don’t look at AI first; look at your business. Where do you want to automate? Where do you already have digitization?
If you start looking at it the other way, from an AI perspective, you’re always going to get hung up on the details and you won’t be able to innovate.
For more information, visit the Canvass Analytics website.