Quebec has a rare opportunity to learn from its past digital mistakes. The harm done to Quebec television by Netflix and Disney+ and the threat from Spotify which weighs on Quebec music are nothing compared to the risk that generative artificial intelligence poses to the Quebec model of health and public education.
That wasn’t his goal. But we’re leaving the special Artificial intelligence decoded with Patrice Roy, broadcast on December 7 by Radio-Canada, with the strange impression that the tens of millions of dollars and the efforts made over the past ten years in Quebec to develop artificial intelligence (AI) have above all served to strengthen the American technological model: we nationalize expenses, we privatize profits.
In the process, we weaken entire industries.
The abandoned citizen
It was Philippe R. Richard, professor of mathematics education at the University of Montreal, who first put his finger on this boho of an AI produced here, but which is anything but Quebecois. “We have this strong impression that the privatization of the system is the latent strategy behind these technologies,” he said, commenting on the Radio-Canada special broadcast.
Like generative AI, the Quebec AI sector may not be aware of its own blind spots. Philippe R. Richard is the author of several works on the role of AI in learning in general, and that of mathematics in particular. He can attest to this: governments have adopted the vision of AI put forward by business people, researchers or scientists who are very competent, but who only seem to see one valid model: the one inspired by the world computer science.
“Institutions seem above all to respond to the needs or vision of the applications industry,” he says. It seems impossible to assert another point of view in the face of this concentration of power and financial resources. Rather than supporting AI models in education that could respond to the development of future citizens, the industry seems to favor approaches mainly focused on private interests. »
“I wonder about the relevance of developing AI outside of collective needs, financed by public funds and guided by issues decided mainly by computer scientists. »
“Move fast and break things”
The generative AI model put forward by OpenAI, Google, Meta and the (rare) others is directly imported from Silicon Valley. It comes down to one maxim: act quickly and break everything in your path. This phrase, often heard in tech subsequently, was formulated in 2014 (and in English, of course: “ Move fast and break things “) by Mark Zuckerberg, big boss of Meta, at the time of Facebook. The latter has become a bit of a pariah in traditional media these days after disrupting part of their advertising model.
To make it more accurate, we must add a sentence to Zuckerberg’s maxim: act quickly and destroy everything in your path… someone will find a way to make money later.
We tend to forget that Netflix and Spotify are not the cause of the setbacks of the Quebec, Canadian, French, German, etc. cultural industries. Netflix and Spotify arrived in the picture a few years after the emergence of peer-to-peer (P2P) file exchange, an Internet phenomenon embodied in the early 2000s by Napster and LimeWire software, then by the BitTorrent protocol.
Netflix, in its early days, mailed DVDs to its subscribers. The entertainment giant we know today simply made its digital shift in time. The same can be said of Facebook and Google. They found a way to make profitable a digital shift in information that until then had mainly been printed or broadcast, but which was already well underway before them.
The next Netflix
These days, it’s AI’s turn to give schools, hospitals and others their “Napster moment.” One difference between the music downloading of the early 2000s and the automated AI generation of school work in 2023 is that today, it is the governments in Quebec and Ottawa that are financing the revolution.
Canada and Quebec, like Europe, are also trying to supervise this revolution to avoid the slippages that are already occurring. In a digital universe where borders do not exist, these excesses give no sign that they will be contained by laws with limited power.
In Canada and Quebec, however, governments have enormous power over AI: its financing. They single-handedly pay for almost all university research and largely fund private research. They could easily make this funding conditional on the research benefiting first and foremost the public networks of schools and hospitals, rather than letting the fruit of this research weaken their very foundations.
“We could ask AI how it can help solve the needs we have in the classroom, rather than confining itself to the needs of industry. But we don’t do it,” laments Philippe R. Richard.
We don’t learn from past mistakes. Coming soon to Netflix: Neighborhood School, seasons one to six…