Artificial intelligence (AI) is rapidly redefining how we interact with data and has permeated most aspects of our digital world—and in more ways than we are even aware of.
How we safely and effectively navigate that new information highway is one of the questions Datacom Director of Artificial Intelligence Lou Compagnone helps answer.
She has her work cut out for her, but then again, as a Sicilian Australian, she says a solid work ethic is in her blood.
Lou shared with me her connection to Sicily, how she got started working in AI, what excites her about AI's future, and how she plans to drive innovation.
What is your connection to Sicily?
My main connection is my father. He was born in Sicily and moved to Australia from Santa Vittoria Domenica when he was about five years old, but he and his family still very much lived like they were in Sicily. The town I grew up in, Donnybrook, in the southwest of Western Australia, is basically the Little Sicily of Australia.
During the fifties, there was a mass migration of Sicilians because of the war. A lot of them ended up in Perth and Fremantle, in particular. A huge number of them actually migrated to Donnybrook.
They almost recreated Sicily in Donnybrook to the point where even their houses looked Italian. They really shaped the landscape. They even grew prickly pears.
So, without even knowing it, I formed a connection with Sicily to the point where when I first visited, when I was in my twenties, I had a sense of déjà vu. I felt like I'd been there before because it looked so much like Donnybrook, and I just really felt at home there.
Sometimes, there's a really inexplicable feeling where you just have a sense of belonging even though it's not somewhere you've been before. When I met my aunt, who lives in Sicily, she said you just feel at home when somewhere is in your blood.
Lou's father was 6'7" and lived in the town's smallest house.
How did your Sicilian roots shape you personally and professionally?
Work ethic is a big thing. So my nonno worked until he was in his nineties. We tried to stop him from working because he was getting a bit too frail to be on our family farm, and we worried he might have a fall. He was always climbing ladders and doing all sorts of things. So we decided one day that we just weren't going to pick him up to take him to work. And then he just started walking on his own. He found a four-wheeler motorbike and just hooned along the road with no license. So we realized we were just going to have to let him work.
My dad was the same. He had that same work ethic and worked harder than anyone I've ever known. He'd always be up at four in the morning and back when it was dark. He worked as an accountant and on our family farm. Even now, he's nearly 80 and retired, but he still does people's tax returns and manages property and stuff like that. So I guess a bit of that has rubbed off on me where I push myself hard in the work that I do, and part of it is because I am passionate about it, and part of it is just this drive that I have just to finish things and do things well. And so that is a bit of a Sicilian thing. Sicilians are sort of battlers. They push through things, and they do it well.
How did you get started working with AI?
I've worked in the tech industry for about 17 years, but my background was actually in service design and futurism. And that might seem like a weird background for AI, but it's helpful. And there are a few reasons for that. If you think of a service as something that helps people to do something, I always think of artificial intelligence as closing the gap between humans and technology more than any technology before. So, it makes it easier for someone to do something. I almost describe it as a beeline. It beelines you between the thing you want, the information you want, or the action you want, and then getting that. Naturally, I always think about closing the distance. Futurism is thinking about what's coming tomorrow and how you prepare for it.
The combination of that—helping people work out what their service needs to be in the future and futurism—probably makes me perfectly placed for it. Artificial intelligence is only as good as the problem that it solves. There's a figure from Harvard Business Review that 80% of AI projects fail. A big part of that is because they haven't actually done service design.
I sort of sweep in and work out what the actual problem is, the real problem to solve. And that works really well with AI.
Describe your role and the group you lead at Datacom.
I'm the director of artificial intelligence for Datacom. I'm mostly an individual contributor because it's basically changing everything that we do. My role is really about operationalizing AI for both our business and customers. The best way of describing it is that a lot of our customers will go to AI summits, or they'll have these tech vendors talking about AI, and they'll be talking about generative AI and context windows and tokens and all this really abstract stuff. And they walk away and go, "What does this mean for my business, and what do I do?"
My role is to help people work out what to do and where to start. I'm doing that for Datacom in terms of working out how we reimagine our services and our solutions, how we build our internal capabilities, how we govern AI (because that's a really important part of making sure that you're doing things safely), and how we have our own innovation pipeline. Then, I help customers do the same. We're very transparent about our own learning journey with AI because it's such a new and changing field. If anyone claims to be an AI expert, I'm like, "Is anyone really?" because the field is changing so quickly.
What challenges do organizations face when integrating AI, and how do you help them overcome them?
We recently did a survey of 200 business leaders in New Zealand. One of the biggest things that came out is that although there's really been an increase, even from last year, in sentiment towards AI and adoption of AI, there's still low governance. Few people have been able to find specific use cases for it. So, the biggest challenge is finding the right problem to solve and the right use case that will have tangible benefits. On the flip side, many people just buy a tool, like they'll buy Copilot, and go, "It's not working." That's because they've just bought this tool without actually working out the use case and the problem to solve.
Many people tell me they feel there's a lot of interest within the organization in people using AI, but it's just happening in lots of different pockets. They describe it as almost out of control as if it is a runaway horse. One of the biggest challenges is how you govern it. How do you coordinate it so that people are doing it in a safe and ethical way, in a scalable way, and in a way where you can actually measure the benefits?
It's not just having a tool for its sake. In the end, AI does use a lot of data. It drives a lot of cloud consumption, so you actually have to do it in a worthwhile way. And sometimes, AI might not be the answer for everything. If you're doing it in a more strategic and governed way, you can have better visibility of whether it's working.
I think security and data readiness are also challenges—making sure people have the data they need to really live up to the ambitions of the solutions they want to build. AI is only as good as the data that it consumes.
We help people with getting their data ready, ensuring that they actually have what they need, and then making sure that they have the security guardrails in place. A big thing with large language models is what happens to your data when using it, where it goes, and what that actually means. I describe to people who are using public AI that they need to understand that anything they type in is going into the big soup of the world. You need to really be cautious about that. And for businesses, it's thinking about actually using enterprise solutions so that you can keep your data close to you.
AI-generated illustration of Sicily
What excites you about the future of AI?
The Institute for the Future recently reported about the different horizons of what's coming with AI. What we see at the moment in Australia and New Zealand (the U.S. is probably quite different) is the first horizon of AI, which has some capabilities but some key limitations. The limitations often include things like truth. We know that AI can hallucinate; if it doesn't actually know the answer, it can just make it up.
A lot of the AI we're seeing can't actually reason. It is really just an elaborate spreadsheet in lots of ways. A very, very good one, a super-smart search. Where it gets interesting and where we'll see those exciting developments is more on horizons two and three. So, on horizon two, we're seeing in places like the U.S. and especially China, an ecosystem of multiple AI models interacting with each other. What you describe as agentic AI is where things can become really special.
It is even more interesting to go beyond the digital to the physical, to interact with the real world through robotics, autonomous vehicles, and neurotechnology.
I monitor signals of what's changing. I'm very obsessed with Neuralink and following how that's going with people who are paralyzed being able to play Mario Kart with their minds. I'm really interested in seeing where that technology will go.
I recently saw something about a talking pet collar, which I thought was hilarious. It's a chatbot for dogs, and it's voice-activated, so it basically gives pets the ability to talk to you.
Then, there's an AI MRI machine that Japanese scientists have invented that records your dreams and gives you the ability to talk to other people in your dreams. I also saw something recently about a scientist who connected a mushroom to a robot body, and it taught itself to walk.
Things like that are interesting to me. If we go back to that thing about the beeline, this goes beyond a shorter distance between two places. This is actually going to unlock things that previously have been behind locked doors. And that's fascinating, whether it's realizing the capabilities of mushrooms—that they might walk if given the right technology—or unlocking the language of pets and what they say to us. The question is whether we should be unlocking that stuff in some cases, but whether we should or not, it gives us the ability to unlock things.
What are your long-term goals for the AI initiatives at Datacom?
I want to build our own internal capabilities. This is a big thing. We already have some amazing people at Datacom. We've got about 7,000 people, and a lot of them are developers with AI skills. Some of the stuff that they come up with is just incredible. Someone's building an evolutionary algorithm at the moment using AI that becomes smarter as time goes on.
I want us to be at the forefront of capabilities because the role that I want us to play with the customers is to help them build their capabilities as well. And in some ways, that's a terrible business model. I want us to help people become independent and make ourselves redundant so that we can simultaneously play a development or tech partner role and a sort of education role.
Part of that is that I want us to learn with our clients. But then the other ambition for me is all around making sure that we play that role in ensuring that AI is safe and ethical and that it's done in a scalable way. And part of that is sustainable because how we design it is really important.
I think in some of the bigger tech companies, the driver is data consumption. You need data to do AI, but the way you design it is key so that it's done where you're actually using data where you need to, and it's all in how you architect it. Playing a role in AI for the good is pretty important to me.
How do you plan to drive innovation and make a lasting impact in your field?
I want to innovate in a human-centered way. I've seen a lot of people talking about how they want to create an AI-driven future. I want to create a human-driven future that's supported by AI. A big part of what I want to do is help humans have agency in the direction this is going and shape the direction. Part of that is actually that you need to lean into it. You need to test the capabilities, and you need to start thinking at a longer horizon.
In futurism, we think about a 10-year horizon. If we start thinking about where things are going, we need to think about the stuff that we both want to enable and are excited about. We want to capitalize on our competitive advantage, but we also want to prevent some of the stuff that we actually don't want to be part of our futures. My plan is to help organizations work out and enable the futures they want and prevent the ones they don't want.
If you enjoyed this article, consider subscribing to my newsletter for more content and updates!