Chance does things funny. UNESCO and Mila – Quebec Institute of Artificial Intelligence are today launching a collective book written by international luminaries in artificial intelligence, and their work takes the form of a warning to all the governments of the planet: act without deadline, otherwise you will quickly be overwhelmed by artificial intelligences (AI) like ChatGPT, a chatbot whose upgrade that occurred last week worries as much as it makes you dream.
The two organizations are taking advantage of the launch of the book, entitled Blind spots of artificial intelligence, to invite Canada CIFAR AI Researcher and Chair Yoshua Bengio and Microsoft Research Senior Scientist Kate Crawford to the stage. The question: how urgent is it for international civil society to equip itself with governance tools adapted to AI? The debate will be interesting between Mr. Bengio, a recognized advocate of a robust ethic in innovation, and Mr.me Crawford, whose employer just fired its ethics committee that was supposed to oversee AI applications within its operations.
Microsoft is rushing to integrate AI tools into its products. In mid-March, the Redmond IT giant introduced Copilot, an interface similar to ChatGPT, which automates the creation of documents in its Word, Excel and PowerPoint software, at the simple request of their users.
“It worries me”, confides from the outset to the Duty Yoshua Bengio. “It sends the wrong signal. I cannot explain this gesture. »
Among the main technological giants, Microsoft has so far presented itself as one of the most ardent defenders of responsible innovation in AI. The mask seems to have fallen.
“Most large companies have made – in appearance at least – a shift in favor of ethics and social impact in recent years. One thing our book does is challenge the not-always-right alignment between appearance and reality,” Bengio sums up.
Take a cue from Canada
The 18 chapters of Mila’s and UNESCO’s book, each written by academic, economic and other luminaries, raise as many questions about the impact that the development of artificial intelligence will have in the coming years, and which remains hard to predict right now. Diversity, culture, education, economy and politics pass through it.
We create powerful tools, which could fall into the hands of those who can afford them, and who could use them to better control populations
One of the fears exposed is that AI will accelerate the concentration of political and economic power in an ever-shrinking number of hands. “We are creating powerful tools, which could fall into the hands of those who have the means to afford them, and who could use them to better control populations”, illustrates Yoshua Bengio.
Hence the importance for governments to intervene, adds Benjamin Prud’homme, co-author and co-editor of the book. According to him, the federal Bill C-27 should lead the way for other countries. Canada is a pioneer in terms of regulating artificial intelligence in particular and innovation in general with this bill, which sets out major principles to be respected by new technologies without imposing specific techniques or means. to be applied to avoid skidding.
The European Union also has a draft framework law for AI, but at the rate things are going, the Canadian law could come into effect before the European law. A good thing, observes Benjamin Prud’homme. “C-27 focuses on risk assessment: the greater the impact of a new technology, the greater the obligations for its creator. It leaves room for innovation while imposing limits. »
The hope with the publication of this book is that other countries will follow Canada and Europe. “It is not enough that only a few countries legislate. There is no need to panic, but it is time to take the subject seriously”, conclude MM. Bengio and Prud’homme.