The state of AI in 2018

By Maxime Ruel
The state of AI in 2018

Artificial intelligence is already firmly rooted in our lives, and rarely does a month pass by without significant breakthroughs making the news. But amidst all the hype, it can be difficult to get a bird’s-eye view of just how far AI has come, the hurdles it faces and the direction in which it’s headed.

In a bid to paint as precise a portrait of the AI sector as possible, The Minutes team scribbled down every noteworthy bit of information mentioned during the AI Forum 2018, a C2 Montréal track dedicated to artificial intelligence and created in partnership with Element AI. Here are the key insights they gathered.

 

The race for intellectual property is heating up

“Montreal and the Canadian AI ecosystems are winning the early innings of the AI game,” says Jean-Nicolas Delage, a partner at Fasken’s Intellectual Property Group. “But it’s a nine-inning game. What if patents are the clutch hitters that allow us to win the game?”

As AI research leads to breakthroughs in every corner of the world, the onus is on innovators to protect their output. “Make sure the money you’re investing is not taking you down a path where you have to pay even more to continue to do your work,” says Pratt & Whitney Canada’s Director and Intellectual Property Counsel, Jeffrey Astle.

Consistently review your business’s innovations to harvest opportunities for filing applications. “These things don’t get to legal organically. Touch base, once in a while, with innovators, review their designs and harvest what’s necessary to protect your products,” Astle adds.

For startups, obtaining patent protection in “key jurisdictions where your technology is proliferating” is key, says Fortress Investment Group’s Managing Director, Intellectual Property Finance Group, Ami Shah. “Two or three European Union countries – or the entire E.U. if you can – China, the United States if you can structure your application properly, and your home country” should suffice – for now.

 

Looking for more post-C2 Montréal goodness?
Check out our Editor-in-Chief’s twelve takeaways from C2 Montréal 2018.

 

Talent shortage threatens to stall AI’s global momentum

“We lack talent in New York, like in many AI hubs around the world,” says NYU Tandon Future Labs Managing Partner Steven Kuyan. Academia’s inability to catch up with market demands partly explains the issue. Of the over 550,000 students in New York City in 2016, only around 4,500 graduated with a computer science degree.

Montreal is home to over 9,000 students enrolled in AI-related programs, but that doesn’t necessarily mean market needs are met. “We have a lot of Ph.Ds, and that’s great. But we need better training at the undergraduate and CEGEP levels for our schools to catch up with market demand,” says Element AI Co-Founder and Montréal Institute for Learning Algorithms (MILA) Director, Yoshua Bengio.

Another factor explaining the AI talent shortage is a brain drain in favour of tech juggernauts. “Top AI researchers are getting poached by big companies, and that’s a problem if their knowledge doesn’t feed back into academia,” Kuyan adds. “Startup nation” Israel experiences something similar. “Large multinational companies soak up all the talent. It’s impossible for startups to compete with the salaries they offer,” says Tel Aviv-based consultant Daniel Singer.

 

Ethical considerations are increasingly under the spotlight

Whether it stems from the data it’s being fed, or how users interact with it, AI bias is inevitable. “But it can, and should, be mitigated,” says Microsoft Office Experiences Senior Designer Ruth Kikin-Gil.

Google Brain researcher Hugo Larochelle shared a succinct checklist. “We need to think of a few things before we start collecting data and designing systems,” he says.

  1. Transparency: Do we want systems that simply make decisions, or do we want systems that explain why they made that decision?
  2. Privacy: Are there concerns that some of the data you train your system on might be leaking towards users?
  3. Robustness: We want systems to not be fooled into making a decision.
  4. Impartiality: Performance has to be the same across different sections of the population.

 

Executives: Keep business value top of mind

“Don’t use AI if you don’t need it,” says Carolina Bessega, Chief Scientific Officer and Co-Founder of Stradigi AI. “When executives call me and say, ‘My CEO told me that we need AI,’ it doesn’t prove they do. If you do need it, choose your battles: start small and build in an iterative way.”

Whenever there’s a promising shiny new thing on executives’ radar, the temptation is high to rush things and skip steps – remember innovation labs? “But why is AI important? At the end of the day, what we want to know is its value for your organization,” Automat.ai CEO and Co-Founder Andy Mauro stresses.
How do we know when AI holds value for your business? Element AI’s Senior Manager of Industry Solutions, Richard Zuroff, has five clues.

  1. When a task or set of tasks rely on a lot of intuition (recognizing a face, for example)
  2. When making the best decision requires different types and significant amounts of data
  3. When it simplifies decision making
  4. When it makes interacting with a system more organic
  5. When it improves a core process, generating a competitive advantage

 

We better gear up for AI-fuelled social change – now!

In collaboration with the AI sector, governments need to consider and prepare for artificial intelligence’s social impacts and draw up legislation to protect the public. Failing to do so would create a swift backlash from the population.

“They will only see the negative aspects of AI, which could very well halt economic progress,” says Yoshua Bengio. We’re also going to need to rethink our social safety net and many programs currently in place to adjust to this new world. “We need to be able to retrain citizens, whatever their age,” he adds.

For Quebec Premier Philippe Couillard, an unequal AI rollout would create its share of problems. “There is a risk of socio-economic polarization – that is between those who have the technology, the knowledge, and those who don’t – and for geographic polarization. We need to include remote regions in this revolution. Otherwise, their citizens will fear for their livelihood, their future and will close themselves off.”