In our opinion, artificial intelligence in the field of Canadian culture should be managed within a governance framework that would be created and administered by people who work in culture. This framework should be flexible, scalable and adaptable to certain cultural communities, particularly indigenous and minority language communities.
We think there should be networks of laboratories across Canada that would look at artificial intelligence issues and allow creators to go out and experiment. These laboratories would also allow them to make decisions about how they want to protect their own data and works. That's not currently the case.
For years, Facebook and Instagram have encouraged artists to publish their works of art on Instagram. These works were extracted entirely by scraping data for use to drive models. So there were lies and theft. Silicon Valley has been doing that for a very long time, as I said in my opening remarks.
So we believe that governance is really at the root of all this. In order to develop governance frameworks, you have to bring people together. We must ensure that there are places where human beings, in this case artists, can come together and make decisions about policies that affect them.
We're not asking the Canadian government to make decisions for artists. We're asking that artists have the autonomy and funding to provide a counterweight to American economic power, which is so vast that it pulls the rug out from under us every time we try to take a step. That's our biggest problem.
Basically, the ArtIA project was funded through a grant from the Quebec government to help us understand the effects of artificial intelligence. We see that all of a sudden the Americans are investing heavily in this area. They do it with capital from Silicon Valley, but they also do it by polluting heavily, consuming natural resources and destroying cultural resources in communities. For example, we see a lot of indigenous art being created by artificial intelligence, by creators who aren't indigenous. So it's a very dangerous cultural appropriation.
As a francophone, I have no artificial intelligence model that represents me. Current models can mimic my language, but they can't understand my cultural specificity. I'm a Quebecker, so it's not too bad, but an Acadian person has a cultural and linguistic specificity that's very different from that of Quebec. Less data on Acadian culture was collected by harvesting, which makes the biases even stronger.
What about a queer artist or a Jamaican-Canadian artist, for example? What about the people who exist at truly complex intersections of our societies, where models, with their biases, end up acting as real homogenizers of culture?
So our position is to let the artists take over, and to do that we would have to invest in those policies.