Plan for Friction: Element AI and the Pursuit of Ethics

“We want to be a contributor for a smooth transition to a more positive future.”

In order for an AI system to be successful, it needs as much data and resources as possible. Researchers and developers may have an outcome in mind, but the idea may change as soon as machine learning algorithms have run their course. That’s the beauty of AI: it can see and do what 10,000-plus humans can not.

The idea behind Element AI began in much the same way. Initially, its goal was to build AI tech and platforms for companies that could not do it themselves. However, as the Montreal-based company began to work with more varied clients—including insurance companies, government-funded media, and massive infrastructure installations—the company’s perspective changed. The original business plan was still sound, but another vision of it emerged, with a much loftier goal.

Element AI found themselves at a point where they could not only contribute to the development of the AI industry; they could lead it as well. Very few companies in the world have this opportunity. Element AI had to act fast, and more importantly, they had to act responsibly.

A Quick Pace

It took Element AI just over eight months to raise $102 million USD in VC funding. This remains as one of the largest funding rounds ever for a Canadian tech company and is a testament to the firm’s goals and its founders. Jean-Francois Gagné, Nicolas Chapados, and Yoshua Bengio started Element AI in late 2016, and that funding round followed in June 2017.

Gagné, now the CEO of Element AI, is a storied entrepreneur, having previously served in executive roles at Planora and JDA Software. He also never went to university. His two co-founders, however, hold PhDs and have helped form humanity’s basic understandings of AI.

“I have spent my entire career with university professors doing technology transfer,” explains Gagné, laughing. “I have probably completed three PhDs in total from all the second-hand knowledge.“

Element AI’s Toronto office.

The history of its founders helps underscore Element AI’s approach. They are committed to the development of AI, both as a community and a technology, but they will always be a business first and foremost, one that creates AI frameworks for some of the largest companies in the world.

“The story of Element AI is a weird one,” says Gagné. “We’re a startup that goes after a scaled opportunity. So our journey is uncommon and unusual. But our vision has been extremely constant, surprisingly. The goal has always been to provide access to tools that are built and designed around AI.”

As that vision progressed, and Element AI found its foothold, Gagné and his team stumbled upon something a bit more meaningful. They wanted to work with partners and create AI “responsibly, with proper governance and frameworks, along with a social responsibility,” Gagné says.

“We felt the only way to respond and go faster than them was through collaboration.”

Only a certain kind of AI company can influence how the sector progresses. After all, niche startups with a handful of researchers don’t really have a say, and experts from the likes of Google’s Deepmind, Facebook and Tesla can dominate headlines with off-handed remarks that offer little value. But due to their founding team and sizeable funding, Element AI realized an opportunity to build a community that could influence the formation and scaling of the AI industry in a positive manner. They were an alternative to big tech, and people would actually listen to them. It didn’t all come at once though.

“It’s hard to answer if the idea of an AI community was a real goal from the beginning,” Gagné says.

“We saw a lot of value in rallying people and sharing, and there are different ways of sharing and connecting and collaboration. Either it’s through open-source, conferences, or partnerships, but one thing we knew from the start is that given the talent, the big actors who are using their weight to dominate the market, we felt the only way to respond and go faster than them was through collaboration.”

For Element AI, collaborating and building an AI community was a natural move, helping them stand out in a big tech ecosystem. For many companies to access AI in a meaningful way, it meant being coerced into opaque and Daedalean agreements—the antitheses of Element AI’s goal for the field’s future.

“Right now, the current position of democratizing AI is having a company work with you only if you put all your data and pay me for a compute cycle on my cloud,” says Gagne. “It’s saying ‘we’re democratizing AI but we’re locking you on our data centre.’ That’s the tradeoff you have to face right now.”

The solution? Open-source the AI world by bringing in like-minded individuals who want to build frameworks based on ethics and responsibility.

“Figuring out the right way to shape outcomes should not be something we think of on top of our ivory tower,” Gagné says. “It should emerge from collaboration and a community that has a voice. The only way we make the right decisions is through adding rational and educated people that are understanding and have access to the tech themselves.”

A Rational Approach

Having a vision is fine, but acting on it is what actually matters. To that end, Element AI is using their historic VC funding to expand their business interests and develop the AI community, through sponsorships, partnerships, and networking opportunities.

They needed to start somewhere, though, and despite the global reach of Element AI, where better than at home in Montreal?

“The only way we make the right decisions is through adding rational and educated people that are understanding and have access to the tech themselves.”

Startupfest, a premier tech conference hosted annually in the city, had its own AIFest, curated and produced by Element AI. This year’s edition featured dozens of founders, researchers and thought leaders exploring the impact of AI on society and business, with discussions on topics like “Exploring AI Through a Social Lens” and “Tackling Gender Bias in AI.”  In the 2017 edition, Element AI sponsored a contest to produce an AI prototype that would fight fake news. The company also produced a three-day AI Forum for C2 Montreal this year, where Gagné and Bengio led keynotes and helped curate discussions on ethics, limitations, and AI for good, among a dozen other topics.

“We brought in all of these experts who the community might not have access to if we didn’t tap them on the shoulders and invite them in,” explains Katy Yam, Element AI’s Marketing and Communications Director.

“What do we get out of it? We give back to the community. It’s hard for us because people see us growing and scaling quickly and that might mean we’re inaccessible. The only way we can be accessible is to be present, be on site, and give people access to us.”

Element AI also has deep relationships with leading research labs, such as the MILA and McGill University’s RLL, allowing them to continually push the boundaries of what it means to be a research-oriented business.

“Our core values are to be a non-predatory player,” says Yam. “It’s to actually open everything up and give access to it all.”

An open space for events at Element AI’s Toronto location.

Like many large tech companies, Element AI is intensely secretive about their work. Very few of their projects are public knowledge, but those are very telling.

The company, for example, worked with the Port of Montreal–which takes in over $41 billion of goods each year, managing over 2,500 trucks per day—to help better manage idle and wait times. The company designed a predictive model that leverages historical and real-time data, giving drivers a better understanding of when to get to the port, hours ahead of time. Not only does this lead to better profit margins for an industry quickly outgrowing its infrastrucutre, it also reduces CO2 emissions and congestion, a best-of-both-worlds solution Element AI has come to be known for.

port of montreal
The Racine and Grain terminals at the Port of Montreal.

Another example is the company’s work with an unnamed insurance giant. Element AI is working to create a predictive satellite imaging model that can better assess potential risks, detect events and then drive action plans. The business upside is evident, but Gagné outlines another purpose once it is further developed.

“It’s AI for good,” says Gagné. “It’s tech we can build that is a by-product of our commercial purposes that can be repackaged and offered to NGOs. It can bring a lot of good and rebalance the power distribution.”

In this case, the technology could help alert NGOs that something might be at risk and then create a plan to help mitigate that risk. Gagné doesn’t go into much detail after that, as it’s clear this is still a developing project. The potential is there though, and it all circles back to the reason Element AI champions community in the first place.

“A lot of the best practices when it gets to governance are aligned with the business interests of driving performance. These should not be mutually exclusive,” says Gagné. “The point here is that, given how profound the transformation over the next 10 to 20 years will be, there’s no way the public will let this go if it reaches a point of imbalance that is too wide.”

“It’s AI for good. It can rebalance the power distribution.”

Element AI also works with clients to ensure proper change management is in place. Their solutions sometimes result in employees being forced to evolve in their roles, and as the agent of that change, Element AI works to make sure everyone is involved in that process, including helping people find other spots in the organization that can drive value. This is the kind of program Gagné wants to see in place when AI comes in to improve workflows.

A Need to Do Better

A lot of people don’t understand the implications AI will have on society. It can factor into weaponry, medicine, the supply chain, banking and more. So who should be the ones to make sure the public is aware of how it will change the world?

“We’ll have some serious decisions to make as a society and citizens,” says Gagné. “People will need to be informed. We always felt we were one of the few that had early access to a lot of these insights, and we had a corporate and personal responsibility to engage with communities, ecosystem, governments and more to provide transparency to what’s going on.”

When someone who is not familiar with the tech world thinks of AI, recent headlines point to a few familiar phrases: killer robots, dangerous autonomous cars, and robots taking our jobs. Technology’s greatest minds do not shy away from the subject, further adding fuel to the topic.

element ai
Gagne speaking on a panel about ethical AI at C2 Montreal.

“I think we should be very careful about artificial intelligence,” Tesla and SpaceX CEO Elon Musk said in 2014. “If I had to guess at what our biggest existential threat is, it’s probably that.”

Other leaders are a bit more optimistic.

“I think you can build things and the world gets better,” Facebook CEO and founder Mark Zuckerberg said in 2017. “But with AI especially, I am really optimistic. And I think people who are naysayers and try to drum up these doomsday scenarios—I just, I don’t understand it. It’s really negative and in some ways, I actually think it is pretty irresponsible.”

Google’s head of AI, John Giannandrea, agrees, saying last year that “I am definitely not worried about the AI apocalypse.”

The fact remains that if there’s still a debate, there’s still work to be done. Element AI is addressing these concerns, both as a business and a community member. The company recently opened an office in London (they also have offices in Toronto, Seoul, and Singapore) to focus on “AI for good.” The aim is to solve global issues by empowering NGOs, intergovernmental entities and local actors; advance methodological research; and promote and connect talented AI scientists in developing nations through fellowships and visiting scholar programs.

Element AI Toronto.

“We want to be a positive contributor for a smooth transition to a more positive future,” says Gagné. “A transition always involves friction and distortion, and this is why we’re so proactive on different levels to motivate the different other actors and communities to rally and actually do something about these frictions because they’re inevitable.”

Gagné and his team have done their part to lead the development of the industry, and are now looking forward. One goal is to find ways to work with partners who don’t have the trove of data needed to feed AI systems. The overarching idea of AI for good, though, will always remain for Element AI.

“Innovation will pursue its way, we want to be in and shaping it to minimize that friction,” says Gagné. “We’re not trying to resist change, not at all. We just believe in responsible change.”