Science and Research Committee on Nov. 14th, 2022

[Valerie Bradford, Liberal, Kitchener South—Hespeler, ON]

This is a bit of a shift of topic.

I understand that you're concerned with the social impact of AI and whether it benefits everyone, and that you have contributed to the Montreal Declaration for a Responsible Development of Artificial Intelligence. Can you tell the committee more about the declaration? What role do ethical considerations have in developing moon shot-type programs more generally?

[Yoshua Bengio]

AI is one technology that is becoming more and more powerful as we develop it more. There are other technologies—like biotechnology, for example—that can have immensely positive uses as we develop them in society, but can also be misused. It's very important for governments to regulate, incentivize or intensify the development that is going to happen to make sure that those efforts work for society in general, for the benefit of democracy, for the benefit of our well-being and so on. That's what was behind the general ideas of the Montreal declaration in going through 10 ethical principles.

If I focus now on the moon shot thing, the important message here is that we can't just leave it alone to the markets to decide what interesting directions are needed. I gave the example of antimicrobials, but I could give examples in the context of the discovery of new materials for carbon capture or better energy storage.

There are incentives—

[The Chair, Liberal, Kirsty Duncan]

Professor Bengio, I'm sorry to interrupt. The worst part of this job is having to interrupt good testimony, but hopefully someone will pick up on that line of questioning.

We will now go to Mr. Blanchette‑Joncas for six minutes.

[Maxime Blanchette-Joncas, Bloc, Rimouski-Neigette—Témiscouata—Les Basques, QC]

Thank you, Madam Chair.

I would like to welcome the witnesses who are joining us for the third hour of this meeting.

Mr. Bengio, it is a pleasure and a privilege to have you as a witness this evening. Allow me to highlight the contribution of your scientific expertise, and also to congratulate you on recently being named as one of the most influential researchers in the world, according to the annual list published by Stanford University in California. It is a source of pride to have you in Quebec, along with the Quebec Artificial Intelligence Institute, Mila, that you created.

The aim of my questions is to determine how the federal government can adopt a vision that is focuses more on moonshot projects. You are an expert in artificial intelligence. Do you think that Canada's vision at present is sufficiently detailed?

As well, you said that we must not let the market dictate priorities, and I would like to hear more from you on that subject.

[Yoshua Bengio]

I think that certain priorities can be and actually are being dictated by the market.

However, because of the market failures that I explained, there are needs that are not being properly met and sectors that are being poorly served. The government can act as a spark plug to get things moving. In some cases, this is urgent. Given climate change or the health problems associated with the pandemics that could be looming in the next few years, we can't wait for things to change in the markets.

I am trying to explain that it is obviously difficult to decide on good moonshot programs. Each organization or company is going to propose its ideas. In my presentation, I did not have time to talk about the importance of international discussions for determining the greatest needs on the planet. I gave the example of climate change, but in what field do we have the greatest need for innovation, and what directions have the most potential for positive change to enable us to face those challenges?

I think we should rely more on a consultation with our international partners and the international organizations that are working on these questions in order to establish the objectives of moonshot programs. We should also rely more on our academics, who are experts in various fields and are a little more neutral than the people with something to sell.

[Maxime Blanchette-Joncas, Bloc, Rimouski-Neigette—Témiscouata—Les Basques, QC]

Thank you, Mr. Bengio.

I want to come back to artificial intelligence, a field you are very familiar with. Quebec is one of the leaders, since it ranks seventh globally in the field of artificial intelligence, according to Tortoise Media's Global AI Index.

Whether the issue is actions or a strategic plan, can the government draw inspiration from certain actions of the government of Quebec, which has decided to invest in artificial intelligence?

[Yoshua Bengio]

I have to correct you: the Pan-Canadian Artificial Intelligence Strategy is an initiative of the federal government. In Quebec, we have been lucky to have a relatively visionary government that decided to double down. So at Mila, the artificial intelligence institute where I work, the investment comes two thirds from the provincial government and one third from the federal government.

What is important isn't that; rather, it is that we work together to determine what missions are most critical. Working with our international partners is also important.

[Maxime Blanchette-Joncas, Bloc, Rimouski-Neigette—Témiscouata—Les Basques, QC]

As you say, the important thing is to work together.

How would you describe the collaboration between the federal government and the government of Quebec in the field you work in, artificial intelligence?

[Yoshua Bengio]

It's quite good. In my opinion, that is not where the problems lie.

There are more problems in fields like health data, where it is obviously very fragmented. There are other problems elsewhere, but I will leave it at that for now.

[Maxime Blanchette-Joncas, Bloc, Rimouski-Neigette—Témiscouata—Les Basques, QC]

Mr. Bengio, I would like to talk about the annual report of the Advisory Council on Artificial Intelligence, which you co‑chair.

We see that Canada is trying to position itself as a world leader in artificial intelligence. However, the planet-wide competition is relatively fierce.

What can we do, in concrete terms? What are Canada's assets and what aspects need to be improved? What form might an artificial intelligence moonshot program take? Are there foreign models we could draw inspiration from?

[Yoshua Bengio]

I think we have incredible opportunities in Canada that enable us to distinguish ourselves even more as compared to what we have already done. I am talking not only about fundamental research, in which we are already a very impressive leader, but also about developing innovations that are going to affect society directly.

There are numerous areas where artificial intelligence can be applied and have an extremely transformative effect. However, the intersection of artificial intelligence and biotechnology has consequences for economic growth and for society, from the point of view of both the economy and health. Our biotechnology and pharmaceutical industry is actually very developed. Our artificial intelligence ecosystem is one of the best. We are finding ourselves at a junction and these two elements could, together, truly change the world.

[The Chair, Liberal, Kirsty Duncan]

Professor Bengio, I am very sorry. Thank you, Monsieur Blanchette-Joncas. You know I aim to be fair, friends.

With that, we're going to go to Mr. Cannings for six minutes, please.

[Richard Cannings, NDP, South Okanagan—West Kootenay, BC]

Thank you, Madam Chair.

Thank you to the witnesses. This is very interesting.

I would like to stay with Mr. Bengio. It's very fascinating to me to hear about how we could develop moon shots. By definition, as you were saying, these are immense projects, very complex, international in scope, and they are tackling the big questions of our time, whether it's health, climate action, etc.

You talked about how we need to decide what the best moon shots are and how this has to be developed with our international partners. Are there any other projects that we've done recently in the world that have shown us the way to do this, how we can decide as a world community to get together and put our forces together to tackle these problems?

[Yoshua Bengio]

I have several examples. The IPCC is doing an amazing job in the area of climate to point out problems and areas that need more effort.

Here in Canada we have an organization that has been extraordinary. It funds moon shot research at a very small scale. That's CIFAR, the Canadian Institute for Advanced Research. Actually, the deep learning origin in Canada happened because of their investment.

The way this works is they put out a call for proposals. Hundreds of groups put out first a short proposal. Then these are evaluated by international scientists in various areas. We have ways of doing these things in the scientific community to compare and evaluate and make these kinds of decisions. Of course, we also need people from governments, from industry, to be part of these discussions, because this is going to be to solve societal problems; but I think there are processes that we already know about to do these sorts of things.

[Richard Cannings, NDP, South Okanagan—West Kootenay, BC]

Once we've decided on what moon shot we're doing, once we've come up with a strategy, a plan, of how we're going to tackle this, of course that will help us find out what further information we need and what data we need to share. I know that in a federation like Canada, data sharing is complex, difficult or impossible at times, let alone around the world.

I come from a background of trying to cross-log data across provinces, and it's very frustrating. When we're talking about these issues, it's even more so.

Then it comes down to the funding as well. Yes, some funding will come from industry if they see a role in it, if they see an advantage for them, but a lot would have to come from various levels of government. Again, are there examples elsewhere in the world that Canada can look at to improve our models of funding?

[Yoshua Bengio]

Let me answer maybe partially by telling you about a particular kind of moon shot that happened recently around what's called a cell atlas. It was about mapping out what different cells in different organisms were doing. This is of interest to both academics and the pharmaceutical industry. By the way, this project was led by somebody who's now heading Genentech, which is one of the big biotechs.

The way this happened was fairly decentralized. There was not just a single source of funding. Scientists from around the world met together and tried to define their thoughts on how they should join forces and what the big questions were, and they drafted proposals for what should be funded and what efforts should be made. Then different organizations and different researchers in different countries went after their funders and convinced them to fund them for these things.

The important thing was that it was coordinated internationally, so all the results were put together and made available to the whole scientific community.

This is interesting. It is more organic, in a sense. Instead of having one big funder that decided everything, it came about because of a consensus of scientists thinking that this was important and that these were the different things that needed to be done. Then individual groups went after their funders—philanthropy or government—to fund different parts of it. This has been an immense success.

[Richard Cannings, NDP, South Okanagan—West Kootenay, BC]

Thank you.

Madam Chair, do I have any time left?

[The Chair, Liberal, Kirsty Duncan]

You have 30 seconds left.

[Richard Cannings, NDP, South Okanagan—West Kootenay, BC]

Quickly, perhaps, Mr. Bengio, could you talk about the transparency that's needed with the public, the trust from the public and getting the public behind these moon shots? I think that's really core to their success.

[Yoshua Bengio]

Yes. I completely agree.

To come back to the previous question about the Montreal declaration, one thing we talked about, which I think is really important for the development of responsible AI and also when we fund expensive research, is transparency—transparency on the process to decide where we invest, transparency on how the work is done, and transparency on the results. This is going to help speed up progress and make sure we don't waste our money.

[The Chair, Liberal, Kirsty Duncan]

Thank you.

Thank you, Mr. Cannings. Thank you again to our witnesses. We're really grateful you're here tonight.

We're now going to go to the five-minute round. I understand Mr. Lobb and Mr. Tochor are sharing.

I will hand it to you, Mr. Lobb.

[Ben Lobb, Conservative, Huron—Bruce, ON]

Thank you, Madam Chair. The first question is for Madam Yeremian.

My question is in regard to moon shots and nuclear reactors. Bruce Power is in my riding, so I'm obviously very familiar with nuclear energy and have been an advocate for years, but here's the problem I see: The moon shot maybe needs to be about how we speed up the environmental assessment approval process.

Bruce Power's site is probably the most studied site in the world over 50 years, and yet I understand that about the fastest you could get an assessment done for a new reactor is probably about eight to 10 years. If we really want to have a moon shot and make progress on the environment and emissions and everything else, wouldn't that be the place to start—to figure out how to do it in two or three? What do you think?

[Rosemary Yeremian]

Absolutely. I'm a big fan of environmental assessments. I don't want to cut corners, but seven years for the Bruce Power environmental assessment was just way too long for a site that has been studied beyond anything that anyone could possibly ever do more.

I was literally, just on Friday, in a meeting with a mining executive in Saskatchewan, who told me that if the environmental assessments are longer than two years, they're just going to go to diesel generation because there's no problem there. It's really a matter of the environmental assessment piece being a competitive disadvantage for Canada.

[Ben Lobb, Conservative, Huron—Bruce, ON]

Thank you.

For Mr. Bengio on Zoom, this goes to what I see in my many years as a member of Parliament, and I know Chair Duncan probably feels the same way: There are issues that have been around for 14 years that are still issues. There are issues about the backlog and getting people processed for immigration applications. There are our port systems; it doesn't matter what end of the coast you're on, there have been issues with the ports forever. We just talked about environmental assessments, and then there's our health care system, homelessness, drug addition. It goes on and on and on, to the point where you can't even put children's Tylenol on the shelves in this country anymore.

Do we need a moon shot in common sense? How do you think we can fix this? We're trying to do the craziest things that would maybe solve society, but we can't even do the simplest of things. What do you think? How can you help us here?

[Yoshua Bengio]

This is a great question. I wish I had answers, but let me try something.

I had a sentence in my pitch about the lack of an innovation culture in everything that has to do with government services. I don't know how to fix this, but Canadians need the services our governments give to be improving and getting better. The kind of culture and investment in R and D that happens in the private sector is not happening so much, not nearly enough, for the things that governments provide. I think this is something that needs to be addressed.