#C2M17: Yoshua Bengio suggests AI innovation still has a long way to go

yoshua bengio

Canada’s first Artificial Intelligence Forum took place at C2 Montreal last week, inviting leading minds in AI research to come together and discuss the AI revolution, which is already impacting the global economy and is poised to grow at an exponential pace.

Presentations, workshops, and master classes were led by Yoshua Bengio, director of the Montreal Institute for Learning Algorithms; Hugo Larochelle, Google Brain group lead in Montreal; Joelle Pineau, associate professor of Computer Science at McGill University; Naveen Rao, VP and GM of the Intel Artificial Intelligence Products Group; and Terah Lyons, former Obama AI policy advisor, among many others.

Novice attendees and industry experts walked away from the event with a clearer understanding of the artificial intelligence landscape — how far we still need to go to get from “weak” to strong artificial intelligence; the massive potential artificial intelligence holds for disrupting just about every industry; and the absolute necessity to start discussing and planning the introduction of these technologies now — both from a level of corporate responsibility, and within our governments.

ai forum

Element AI, the world’s largest AI research lab, approached C2 to run the forum as they saw this as an opportunity to reach a broad audience and bring together major players in Canadian and international AI research communities. Keynote speeches, which were accessible to all attendees of C2, were led by Bengio and Jean-François Gagné, co-founders of Element AI, as well as Rao.

“Part of our mission is to make sure that we promote and help people go from fear of AI and misunderstanding to being favourable or preoccupied,” said Gagné when asked about the intention behind running the AI Forum. “For us, C2 was the perfect place to do that as they attract a lot of people … it gets people from all horizons to connect, debate and hopefully get them to appropriate the technology and now be more knowledgeable about it.”

In his Master Class, Bengio stressed the importance of rethinking our social systems to deal with the changes that are coming, although he — and many of his colleagues — also noted that the progress we are hearing about in the media is, in many cases, exaggerated.

“All of the industrial successes that we have now rely, for the most part, on supervised learning. We have to tell a computer what the answer should be for each example,” Bengio said. “If you look at the mistakes that these systems make, you realize that they still learn about the world in a very superficial way, that they understand things that are similar to what they’ve been trained on. But if you go in a direction that’s different, they make a lot of mistakes.”

Naveen Rao
Photo by Myriam Bariltessier

He also described areas where we have made great strides with AI technology, which he sees as a means to help more people and create massive economic value.

“We’re opening a door to the treasure of knowledge to hundreds of millions of people,” said Bengio, noting applications of AI including customer service by using text or speaking chatbots, numerous uses in healthcare, systems that can provide information from manuals, dialogue systems for entertainment and education, and legal services.

“A large fraction of the work could be done by computers much more efficiently, so we’ll need less people,” he continued. “Of course, in terms of economic value, this will be huge…you have huge value and of course we want to be among the organizations and countries that take advantage of that value. At the same time there’s going to be a tremendous disruption in job markets so government will have to manage that transition.”

Lyons, former Obama AI policy advisor, also noted that “technology is not destiny,” and that policy makers and public servants have a “massive obligation to actually craft smart policy around mitigating the challenging effects of AI”.

“The next challenge we’re facing is an education challenge,” continued Gagné. “To be a global leader, not only do we [in Canada] need to have the technology, but we also need to make sure that the population gets comfortable with the technology and that we make progressive legislation happen.”

While covering the impressive successes we have already seen in AI language and image recognition, translation, and simulated environments, Gagné, Bengio, and numerous other presenters also stressed the importance of Canada’s position, not only as the leading AI research centre, but also as a leader in the implementation of these technologies.

“Our biggest challenge as a society is to make sure that the big transformation that we’re going to go through — and is going to hurt people — is going to generate value,” said Gagné. “We need to make sure that where the value is going to be created, is at the same place where the taxes applied on that revenue will be able to protect the people who are going to be disrupted. And that’s a huge challenge. We’ll try to play a role, but we’re only one.”

Feature photo by Allen McEachern

Avatar

Lauren Jane Heller

Lauren Jane Heller is passionate writer and storyteller. With a background in documentary film and journalism, she has now found her niche writing for and about the continually evolving world of technology.

0 replies on “#C2M17: Yoshua Bengio suggests AI innovation still has a long way to go”