I have up to seven minutes.
Thank you, Mr. Chair.
My thanks to all the members of the committee.
My name is Paul Gagnon. I am a legal advisor with Element AI. I deal with matters of intellectual property and data.
I'll give the testimony on behalf of Element AI in English, although I welcome questions in French in the later stages of today's hearing.
Thank you for giving us the opportunity to speak today. Element AI is a Montreal- and Toronto-based artificial intelligence product company. We're celebrating two years of activity shortly and making headlines all across the world with offices in London, Singapore and Seoul.
It's a great privilege to come before you today, especially as an IP geek. It's a great occasion to come in and comment on copyright reform.
Element AI is bringing fundamental research as quickly as possible into actionable products and solutions for companies. Our momentum is favourable. We're quite proud of our achievements, but the best is yet to come, which is why we're here. We want to discuss how Canada's place as a world leader in AI is not guaranteed. We'd like to invite limited and targeted reform within the Copyright Act, in order to clarify a specific use case around informational analysis, also known as text and data mining.
This lack of certainty around the act impacts a very important activity for the development of artificial intelligence. This informational analysis, within the context of fair dealing, would be beneficial for all Canadians and more specifically as well, those who are active in the sector of AI.
This targeted exemption would help us secure a predictable environment for AI, in order for it to maintain its unprecedented growth. Competition in AI is global. Other countries are actively building policy tools to draw in investment and talent. We urge Canada to do the same.
When we speak to informational analysis, what are we referring to? We're talking about analysis that can be made of data and copyrighted works, in order to draw inferences, patterns and insights. This is informational analysis, not the use of the works themselves, to draw from them and use and extract information from these works. It's distinct from using the works themselves. It's about abstraction. It's not about commercializing the works themselves and undercutting Canadian rights holders.
As a quick example, if we were to look at the paintings in this building when we walked in, it's not about taking pictures of these paintings and making T-shirts. It's about looking at the paintings themselves and drawing patterns, measuring distances and measuring the colours and tones that are used by artists.
To use another example which is more relevant to your day-to-day work in Parliament, if you were to look at the Hansard debates and use them for informational analysis, we wouldn't be binding books and selling the debates. Perhaps we'd be using translated words to build more functional algorithms to translate works. We see that here there's an abstraction. It's not the work itself; it's the information that we can derive that's used.
Data is truly the fuel that powers the engine of AI. Algorithms in AI-based products need diverse, representative and quality data. That is the supply chain around AI to provide actionable insights and data, in order to provide better products and services.
A good old expression in computer science is garbage in, garbage out. This truly applies to artificial intelligence and informational analysis. Our AI will only be as good as the data we provide to it. Therefore, the targeted exemption we want to speak about today aims to broaden this scope, in order for our AI to be quality, representative and in turn, made accessible for Canadians everywhere.
We think that with a clearer right and resolving the legal uncertainty around informational analysis, we can drive fairness, accessibility and inclusion of AI-based solutions. Really, better data means better AI.
Under the current Copyright Act, how is informational analysis understood? How is it apprehended? The Copyright Act protects copyrighted works, but it also protects compilations of copyrighted works and also compilations of data. There are three fronts that are protected.
As you've seen in the works of the standing committee, the Copyright Act is about balancing different interests, users' rights and access, but also rights holders, which is why we suggest that informational analysis be made part of the fair dealing exemptions.
Fair dealing exemptions are limited in purpose. The act clearly states the purposes around what kind of intention we can bring to analysis we can draw, for example, research use, private study or news reporting. Where there's an overarching public interest, the act has clarified that there's a clear purpose that's permitted within fair dealing.
Informational analysis, as we've explained it, how is it apprehended as the act exists today? It's not clearly addressed, and so there's legal uncertainty around this. We could look perhaps to the temporary reproduction exemption, but that doesn't quite fit. If you turn to specific fair dealing exemptions, research, private use, this isn't clear. We think it's within reach of Parliament to clarify this, and in turn help drive investment and certainty for Canada's AI sector.
If we look at research purposes more specifically, the uncertainty here is quite impactful. Indeed, it could permit the informational analysis itself under research, but there's clear uncertainty as to whether we can leverage this research into products and solutions. Relying on solely the research exemption might not be enough.
We suggest this exemption not be limited to the identity of the specific entities conducting this informational analysis. Truly, if you look at research around AI, the public sector is quite active, as is the private sector. At Element AI, we collaborate every day with researchers at universities across Canada. If we were to clearly exclude commercial entities such as ours to perform this research, it would fundamentally misapprehend the nature of research in Canada.
What's the impact of this uncertainty? In time, if we do not bring this additional added clarity to informational analysis, it can have real and practical impacts on Canada's competitiveness in the AI sector. It can deter R and D investments, and create risks for businesses. Not having legal clarity around informational analysis disproportionately impacts startups and SMEs. Why? As we all know, certainty and predictability are the currency for our entrepreneurs. On the other hand, big players have deep pockets to litigate and fight through this uncertainty. We might not have this luck for our startups and SMEs.
Should we wait for this to be litigated in court and clarified downstream two or three years on? We don't think so. There's a great opportunity to clarify this now.
We often hear that we live in the age of big data, which is true. In terms of volume, there are massive amounts of data generated every day, but there's a huge data gap. Because we generate that data, it doesn't make it accessible to smaller players. Indeed, there's a huge gap between who controls and has access to this data. To ensure the competitiveness of our SMEs, our startups and our more established companies, it's essential to make sure there can be clearer access to this data in order to bridge this data gap in order for this chasm between Internet giants not to be broader.
Why informational analysis fits well under the fair dealing exemption is it benefits from past interpretation of the courts. It's a clear framework, and it's one that aims for fairness. Through case law, it has established clear criteria we can rely upon. Those criteria are flexible and adaptable to different use cases.
Fundamentally, copyright protects the expression of ideas and information, not information and ideas as such. It's a fundamental principle of copyright law. Fair dealing, in this context, especially for informational analysis, brings a clear fence that is quite reasonable, and clearly brings more certainty to our private sector.
The key point here is that an exemption for informational analysis can help democratize access to data and create certainty for the emerging AI industry. In turn, this will help maintain Canada's leadership role as a global hub for AI.
Thank you.