(Montreal) Ottawa is betting on self-regulation in terms of artificial intelligence, while waiting for more robust laws to be adopted to protect the population.
The Minister of Innovation, Science and Industry, François-Philippe Champagne, presented Wednesday in Montreal, at the opening of the ALL IN conference on artificial intelligence (AI), the “Voluntary Code of Conduct aimed at responsible development and management of advanced generative AI systems”.
This code, which was to be signed by several large companies working in the fields of IT and AI, commits the signatories to adopt several measures “to ensure that risks are well identified and mitigated”.
The idea of asking a private sector industry to self-regulate is not new and has rarely been successful, but Minister Champagne could hardly be blamed for lacking enthusiasm for his code of conduct.
I think that the community will still be able to self-regulate because someone who signed it, given the expertise we have here, I think that bad practices would be highlighted.
François-Philippe Champagne, Minister of Innovation, Science and Industry
“I can’t see someone signing it and not respecting it. I think that at that point, users, customers and customers would react quite strongly,” he said.
Malicious use
The code thus highlights certain risks of malicious use, for example “the ability to produce realistic images and videos or to impersonate real people can enable deception on a scale that can harm important institutions , in particular to the democratic and criminal justice systems. It also raises significant risks for individual privacy rights.
The public knows these risks, hence the distrust that François-Philippe Champagne hopes to alleviate. “We have reached the point where we need to take action. Clear frameworks are necessary to restore confidence. »
The signatories thus undertake to adopt accountability measures involving risk management and security measures to ensure that the operation of the systems is safe.
It also involves ensuring respect for justice and equity in the deployment of systems and transparency “to enable consumers to make informed decisions and experts to assess whether risks have been adequately managed”.
Minister Champagne believes that this issue of transparency is fundamental so that consumers know what they are dealing with. “If you apply to have a loan, an insurance policy, or even a job, we are not too sure today: was it a human who decided or if it was an algorithm? And did this algorithm choose your postal code, your age? Transparency will restore consumer confidence,” he believes.
The code also requires the existence of human monitoring of systems and their securing against cyberattacks, among other things.
Waiting for C-27
The importance of ensuring ethical, responsible and prudent development of AI was underlined by all of the numerous speakers at the opening of the conference, in particular the mayor of Montreal, Valérie Plante, for whom innovation must be “in phase with our shared values of ethics, fairness and accountability”.
Following suit, the Quebec Minister of Economy, Innovation and Energy, Pierre Fitzgibbon, argued for the need to “collectively reflect on the issues and raise awareness among all stakeholders of the importance of establishing benchmarks clear rules for responsible and prudent development of AI.
François-Philippe Champagne assured that Bill C-27 will partially respond to concerns, but he recognized that “the law will take time, but people want us to take action now to ensure that we have specific steps that companies can take to build trust in their AI products.”
The scientific director and founder of MILA, Yoshua Bengio, insisted that it is “really important that we make the investments and the efforts in all organizations to protect the public, to ensure that artificial intelligence will work for the general good of society and we do it with our international partners because artificial intelligence has no borders. Beneficial uses as well as dangerous uses can come from any country in the world,” he recalled.
The example that François-Philippe Champagne regularly mentions is that of cloning.
“At the time, humanity came together to set markers. We decided we wouldn’t clone humans. Perhaps scientifically it is possible, but humanity has given itself rules and safeguards and that is exactly what we are doing today.
“Artificial intelligence, yes, for responsible innovation, but we must also give ourselves a framework in which we say: humanity will say that there are lines that must not be crossed to protect individuals. »
On Wednesday morning, a dozen companies working in the field had signed the code of conduct, including one representing around a hundred “startups”. The conference welcomes some 1,400 participants on site at the Palais des congrès de Montréal, and more than 10,000 participants online. On Wednesday and Thursday, a range of experts from industry, research, academia, the public sector and others from around twenty countries will discuss artificial intelligence in a context where it is evolving at a rapid pace. unforeseen and takes hold in all spheres of activity.