We spoke to Remi Duquette, Vice President of Innovation and Industrial AI at Maya HTT. We talked about what the training of engineers will look like in the future. How should future engineers be educated? Do we need specialized artificial intelligence or generalized humans?

Well, eventually we're going to talk about the engineer of the future, but I wanted to start off by talking about the engineer of the past. Can you tell us a bit about your background and how you got here? Well, my background is probably not atypical in terms of engineering. In the 90s I did my engineering degrees and graduated from McGill and University of Toronto, and then went into aerospace engineering. That really was my background. Stayed, of course, for probably about a decade in aerospace engineering, and then moved on to the software engineering world and developed all sorts of funky and fun applications. Now I'm in charge of innovation at Maya HTT, and we develop multiple software solutions for different engineering domains not just space, although we did start in space but we now are developing software for about 12 to 15 different industries with experts in each of them. In the last decade, I really moved on to AI and machine learning as a practice within our company.

And it's changing at a really, really rapid pace, isn't it? Whole swathes of existing skills, like learning to drive, are likely to disappear because they're going to be taken over by AI. Is that what you see is or going to happen? Well, certainly as you mentioned there, there are some skills that will be made obsolete by some of the new technologies that are emerging. Some are a little bit scary and we'll have to see how they evolve and if it's as rapid as we think it will be, but certainly there will be some rapid change. My nine year old is at a summer camp coding today and it's like, "Well, I didn't have a computer at her age because computers were starting out but not really widespread at the time."

So it's kind of an interesting thought to see, and how rapidly these changes will occur. But yes, I'm certainly bracing for a lot more of those changes and that's why I think the future of engineering is important and a really critical topic to address because the engineers we train today... We can't just overspecialize them on specific technologies as we know those may not be available in five years' time because they keep on changing.

How do you think AI is going to affect future engineers' jobs? AI is really a technology, a new way of dealing with data and learning from data. Engineers have used data in the past whether it's in controllers, in the manufacturing space, or in operations so it's not really a new topic. It's been made a lot more powerful by the computers that we have and the amount of data we have at our fingertips and are able to process, whether it's from telemetry, real-time telemetry, or additional sources that we can tap into.

I think AI will change engineers in a couple ways. One, in augmenting them and two augmenting their capability. It’s going to change the way that we think of a design cycle in engineering, or product design.

Nowadays, that product sends back telemetry back home, so to speak, and tells you new things in the environment that you may or may not have put into your design in the first place. It brings new ways to think about how to intuitively design and put forward some interesting new ways of coming up with amazing new products that we couldn't conceive before. That kind of feedback loop that's a lot more rapid and real-time, gives us engineers more tools and interesting information to process.

Do we think that the engineering graduates today are having the correct training to play a part in this future of engineering? Well, certainly, and I've been in many interesting discussions and conferences with a lot of people that are teaching our engineers. In the past the focus was really more in problem solving skills and I'm going to call them mathematical skills, technical knowledge and logical reasoning and thinking. As, we look to the future, we're at a crossroad where we grapple with generalization versus specialization.

In a way, an analogy to AI. If you over-train and overspecialize a little model, well, at some point it just does not generalize very well and it can't adapt very well. It's the same thing for training engineers.

If I had told you that we would be contemplating self-driving cars 10 years ago, you would've laughed at me but now we're getting closer and closer to that reality and people are not laughing anymore, and they're investing a significant amount to make it happen. That's really, I guess, the point there on specialization versus generalization of engineers. It’s definitely moving from pure mathematics to really adaptive skills that will make you a really good engineer that is able to evolve with the pace of technology and adapt with new technologies as new tools.

If you over-train and overspecialize a little model, well, at some point it just does not generalize very well and it can't adapt very well. It's the same thing for training engineers.”

Remi Duquette, Vice president of Innovation and Industrial AI, Maya HTT

There is an argument to say that there is a problem with the sort of engineering current software engineers do, in that you spend your whole career trying to drive the software as much as anything else, and not enough time doing real engineering. Assuming that AI is going to free engineers up to do proper engineering, making decisions, giving insight and not just be “mesh monkeys”; Is that how you see the future going? I definitely see that as a trend, and certainly AI as a technology does bring those insights that kind of bubble to the surface, those insights that may have been hidden in the data or in the software. Instead of having engineers and humans going through and sifting all of this, they can employ the idea of generative AI programs that will give you the best couple of solutions and then you will apply your engineering judgment to pick the right one. I mean, it's still going to be a probability game where AI brings about what's most probable, and people need to think in a different way in those environments. I do see that trend certainly increasing in the future.

And while we will get some brilliant solutions we will also see some completely unfeasible solutions. We still rely on engineers to spot the things that are completely unfeasible. It’s not just unfeasible but dangerous sometimes. You see it in all sorts of things and that's why AI needs to be understood and harnessed in the proper way. AI, again, is purely and simply a new tool in terms of its power. It's been there in terms of algorithms for machine learning and deep learning has been there for over two decades, in which time we've seen it evolve in some brilliant ways, and not so in others. For example if you look at the data used by social media platforms to train chatbots, it may be deemed as unethical in the way that it uses language.

And not forgetting, of course, that human engineers often make bad decisions, and sometimes you're going to need AI to pick up those decisions, as well.

When we talk about the engineers of the future, actually those are the engineers that we're training today because if you graduate next year, basically you're still going to be working in 2060 or maybe even 2070, by which time the world will have changed completely. We have to start teaching these skills, don't we? I guess it happens naturally, but the future starts now. It does start now, and actually it starts with even us, you and me. I mean, I graduated two decades ago now, but I keep on learning. Every year I make a point of learning, whether it's small or big, a new skill to put in my arsenal of skills. I hate to kind of quote someone like Einstein but once you stop learning, you start dying.

Further reading

To listen to the full interview, download the Engineer Innovation podcast