Isaac founder talks AI, data privacy, and more

There’s telematics data, data everywhere, and at times it can seem like fleet managers are drinking from a fire hose. Identifying trends and offering true business insights will increasingly rely on artificial intelligence (AI) and machine learning (ML). Jacques DeLarochelliere, the president of Isaac Instruments – a business built on generating data – knows such technology will play an important role in the future of trucking. Today’s Trucking sat down with him to discuss his thoughts on AI and ML, and where they will go from here.

Isaac president and co-founder Jacques DeLarochelliere: Any data any breach could affect the valuation of your company. (File photo)

Today’s Trucking — You’ve expressed some concern that “artificial intelligence” and “machine learning” are sometimes misapplied by overly zealous marketing folks.

Just to clarify things, how do you define the difference between artificial intelligence, machine learning, and what you often refer to as just clever engineering. Jacques DeLarochelliere — Every buzzword that becomes clickbait is very tempting, compelling, for marketing or even sales. At one point, [the buzzwords were] Y2K, the internet, dot-com, web, the cloud, whatever you want to name it, gamification, big data, blockchain, IoT.

We’ve heard all of that. The most recent ones are AI and ML — machine learning. Machine learning is sort of a subset of artificial intelligence.

You need a computer to learn. The algorithms are improved automatically. So, you throw good examples and bad examples at the computer, and it’s going to come up with the algorithm.

There’s no one actually coding that. [With AI], humans have to understand and do the analysis and predict the conclusion. Today’s Trucking — You’re a champion of privacy considerations. Are we limiting some of the gains that can be realized through artificial intelligence and machine learning by being so protective of the data?

Should we open things up a little more. Jacques DeLarochelliere — What does matter is who holds the data… It’s really important to understand the value of data, and trucking fleets should know where their data goes.

If you’re trying to feed the beast, to make it improve, then you’re sharing large amounts of data — all of it if you can — because the larger [the] amount of data, the better the algorithm is going to get. The machine learning is better. But your business is your activity and your assets.

So, if you’re a CEO of a trucking company, your responsibility toward the shareholders is to protect that valuation. But you’re going to let go of the data … to help the algorithm, which may be helping your competitors more. When people are signing up for some of the systems, some of these offerings, [the data sharing is] part of the privacy rules that they’re clicking off on.

They might want to take a look at what exactly is being shared. There’s many applications of ML and AI that could still protect data. I’m going to give you an example: The Isaac Coach’s simple gauge that the driver is using to save fuel.

We use a large amount of data for all kinds of different cases. But to make that specific use in the truck, we use this driver’s data and the truck’s data. Nothing else.

It benefits a single individual. This individual is not sharing the data with the community, but is getting the benefit of what we could extract [from] this huge database. Today’s Trucking — Can you offer some other examples of the way AI and ML could be used in trucking?

Jacques DeLarochelliere — Routing. In every trucking fleet, there’s a crucial member of the dispatch team that knows everything about every route. You’d be surprised to know even the large fleets still have a champion that gets it all figured out and knows where truckers like to stop.

It’s crazy. Too much of it to this day is still done by humans. We could share and learn from this, but this implies that you’re sharing data.

I’ve got to know that your trucks are blocked in traffic in such a place to reroute somewhere else. That’s a very basic example — but you pay the price, and I benefit from your loss, right? You don’t need AI or machine learning to do that.

But there’s extensions of that, that will benefit from machine learning and AI. Today’s Trucking — Were there some particular enablers that actually make all this possible? We’ve seen announcements about massive data warehouses.

There’s the connectivity between the fleet office and the vehicle. And now the broader use of sensors on the truck. Is there any one thing in particular that’s really opened the door to AI and ML?

Jacques DeLarochelliere — Yes, APIs — Application Programming Interfaces that allow systems to talk to one another. Telematics providers such as Isaac, we are – I’d say – the first in the food chain. We are raw data provided to this food chain.

We collect the actions of the drivers in the truck, and the reactions of the truck, and provide them to the back-office software. There’s multiple applications we talk to. You need to share data.

API’s have allowed this ecosystem to feed data more than you know, much more than it was five 10 years ago, so this data allows us to draw conclusions that we could not draw before. Of course, even without machine learning or AI, we can still have the same conclusion. It takes more time.

Today’s Trucking — It would just require a massive spreadsheet to gather an in-depth answer … Jacques DeLarochelliere  — … and maybe you’ll never find it. If you go into a computer, you’ll have to look at a broader set of data.

And if we have more applications bringing data there, you might find better solutions in places that you didn’t expect to find them. Today’s Trucking — We’re talking a lot about possibilities. Let’s talk about the limitations of artificial intelligence and machine learning as it exists today.

Are there technological limitations, or is it really just a matter of feeding today’s systems with more data? Jacques DeLarochelliere  — The human factor is going to be the limitation. I mean, you’ve got to find people developing these technologies.

It’s something to find the talent and get them work together and aligned on a common goal. It is not easy. Machine learning, AI, is not something that you learn during the first year of an engineering degree.

It requires quite a bit of instruction and education Today’s Trucking — Where do you expect AI and ML to go from here? Are there particular trucking-specific processes or questions that you look at and go, “That’s really where we’re going to start seeing some differences”?

Jacques DeLarochelliere  — One of them is to keep the wheels turning.

By how much of a factor can we reduce the downtime, or the stop time of attracting and waiting for a load, or waiting for parking, or rest areas?

  • This interview has been edited for the purposes of clarity and length.