ChatGPT, the conversational robot developed by the firm OpenAI, has just made enemies, and not just any. Last week, hundreds of tech figures called for the temporary suspension of development of certain features of advanced artificial intelligence (AI) systems.
The letter is signed in particular by the world authority of AI, the Montrealer Yoshua Bengio, the entrepreneur Elon Musk and the co-founder of Apple, Steve Wozniak. Their purpose is broader than ChatGPT, of course. They worry about the “great risk to humanity” posed by the development of ever more powerful systems that can rival human intelligence, with the inherent risk of propaganda and misinformation. As alarming as this public outing is, the remedy proposed by the committee of wise men is rather expeditious given the slow pace of legislation.
A moratorium of at least six months on the development of systems better than GPT4 will do little to delay the frantic planetary race to implant AI in our lives. We are already at the stage where the creators of these systems can no longer “understand, predict, or reliably control” their creatures, said the signatories of the letter. If the scientists behind these tools no longer keep up with the pace of progress, imagine how much public decision-makers are left behind.
In democratic regimes, it is on them that we will have to count to mark out this unprecedented field of knowledge and to ensure that AI is used to improve the well-being of citizens. Canada and the European Union (EU) are looking for a way to achieve this, in a risk management approach. The Federal Minister for Innovation, Science and Industry, François-Philippe Champagne, has also welcomed the open letter from personalities from the world of technology.
With Bill C-27, Ottawa is preparing to regulate AI in an innovative way. The Artificial Intelligence and Data Act (LIAD) proposes striking the balance between supporting AI research and innovation and proactively identifying and mitigating risks to prevent harm and discrimination.
With its 850 start-ups, 20 public AI research labs, 75 business accelerators and 60 investor groups, Canada (and especially Quebec) is a world leader in AI. The country is well positioned to set the tone in public policy. The federal approach is very promising and, if the United States does not oppose its protectionist reflex in favor of the American technology giants, it has a chance to spread.
However, we deplore a void in this reform project, and that is the absence of reflection on respect for intellectual property and copyright. ChatGPT like its peers have scoured the Internet in search of content to enrich the neurons of their digital brains. In the United States, press publishers understand that they have been duped, once again, by the collection of massive data on their platforms, a necessary step to enrich and improve the algorithm of conversational robots. Some publishers are considering demanding compensation from Microsoft and Google, the main AI service providers. Publishers are certainly not the only ones to have unknowingly fed these almighty robots with their content. It is not too late, in Canada, to broaden the reflection in this regard.