Mr. Chair and honourable members, I appreciate very much the opportunity to speak to you this morning.
My organization, the Scale AI supercluster, is the only truly national AI organization in Canada. We've supported over a hundred AI industry projects from coast to coast, providing us with unique insights into the business side of Canadian AI.
In 1933, Franklin D. Roosevelt stood before his nation and said, “the only thing we have to fear is fear itself—nameless, unreasoning, unjustified terror which paralyzes needed efforts to convert retreat into advance.” I think FDR's words echo here today. In the natural world, fear is survival, but in the human world, we balance fear with reason. The past year has seen incredible advances in AI progress. Fear is understandable, but it can't govern our response.
Here are five things that I think our rational minds needs to keep close. The first is that AI is still just math that makes predictions based on data, even ChatGPT, even Stable Diffusion. When progress catches us by surprise, we must adapt.
For example, we've known forever that photos, audio and video can be altered and even faked. Hollywood CGI even now seems to bring dead actors back to life. The first time we saw deepfakes, yes, we were shocked, but now it's time to adapt. If Joe Biden gives you a robocall later tonight, Mr. Chair, I don't think you're going to be fooled. Fool me once, shame on you; fool me twice, shame on me.
Second, we know that fear leads to poor decisions. History shows us regrettable legislative decisions motivated by fear of political belief, of race, of orientation. Fear is not a worthy legislator. For example, we know that AI doesn't create discrimination. Bias is a human failing. It shows up in our data. It shows up in our processes. Nevertheless, bias and ethical AI are now the subject of important discussion and research within the Canadian AI community, because these are also Canadians who care about their privacy, their human rights and their children, just as we do. As Canadians, we are much better off having our fellow Canadians working on these problems rather than hoping foreign companies will solve them in line with our values.
The third thing to know about Canadian business is that it needs Canadian AI. We lead in AI research, but Canadian businesses' adoption of AI has been slow. Scale AI is working to correct this because it's essential to narrowing Canada's famous productivity gap. We're supporting everything from predicting consumer demand to keeping trucks on the road and our roads in good shape to improve Canadian productivity. Avi Goldfarb of the Rotman School of Management sat here and said that, if Canadian businesses do not continue to adopt AI, the standard of living of all Canadians will be affected.
The fourth thing to stay rational about is that we can't regulate the unknown. Vague or overly broad rules won't protect Canadians, but they will force Canadian AI companies to go elsewhere to find predictability, leaving us in foreign hands. That's not better. I have some examples to share if anyone wants. However, we must stay true to our legislators' instincts. Clearly defined, known or foreseeable things create clear requirements with a mechanism to adapt when the unknown finally arrives.
The last thing that we must recognize is that this committee has had almost no guidance from AI businesses. They've made up just 5% of your witnesses. Shouldn't we give a voice to those who are affected?
The joining of AIDA with parts 1 and 2 hasn't helped, and splitting them apart for a proper study makes sense. The Canadian AI CEOs that you have heard from say that they are ready for regulation, but they were also unanimous that the legislation must be clear and internationally consistent.
JF Gagné sat here and said that, if the goal is to have a framework that actually works, then it's important to ensure that it's not overly general because that makes it difficult for Canadians to innovate or have confidence in their regulatory framework. Business confidence in regulation can be an advantage. In the fifties, despite intense fear around nuclear technology, clear Canadian regulation fostered the development of the ultra-safe CANDU reactor, a competitive advantage for Canada that sold around the world.
As I said, international alignment is essential, but the EU and U.S. positions are not yet clear. President Biden has essentially only ordered his departments to study and report back, and a key EU member state has now already signalled that it doesn't support all of the EU act.
How can we align at this moment? Some want us to hurry AIDA forward, alleging that there's an AI wild west, but that's the fear talking. Existing laws apply to AI just as everything else. There is no legal vacuum. We have time to make it clear and to make it consistent.
In closing, I encourage you to respect your fears but not to let them legislate. Canada needs Canadian AI. Canadian AI needs clear and consistent legislation, and that needs guidance from Canadian AI businesses.
Thank you very much for your time.