Geoffrey Hinton at the University of Toronto on Wednesday. The Nobel Prize winner says he will stress to Canada's Artificial Intelligence Minister the importance of regulating AI against what he sees as the dangers of the technology.Fred Lum/The Globe and Mail
Nobel laureate Geoffrey Hinton said he will tell the federal Minister of Artificial Intelligence that Canada needs to regulate the technology when the two meet on Thursday.
An emeritus professor at the University of Toronto, Prof. Hinton spoke briefly with The Globe and Mail after a talk on Wednesday and said he has a meeting scheduled with AI and digital innovation minister Evan Solomon.
“This stuff needs regulating,” Prof. Hinton said. “He’ll get a lot of pressure from industry not to do it, but it’s potentially very dangerous.”
Mr. Solomon has said the government is not abandoning AI legislation but it will not reintroduce the Artificial Intelligence and Data Act, which died when Parliament was prorogued in January.
The act aimed to put guardrails and enforcement measures around “high-impact” AI systems, including algorithms that make determinations related to employment and health care.
British-Canadian Geoffrey Hinton and U.S. scientist John Hopfield, winners of the 2024 Nobel Prize in Physics, share their concerns about the artificial intelligence boom.
Reuters
Mr. Solomon has also said Canada is moving from “over-indexing” on regulation toward more adoption.
“Companies will always say that regulation will interfere with innovation,” Prof. Hinton said.
“There’s a trade-off. The big problem is that unless you can get international agreements, countries that don’t regulate will have an advantage over countries that do. That’s the same for exploiting natural resources.”
The public isn’t aware of the risks posed by AI, he continued.
“I see my role as educating the public,” he said.
There has been a global shift of sorts away from regulating AI to instead rapidly deploying and adopting the technology for economic gains.
U.S. Vice-President JD Vance summed up the new attitude in a speech at an AI summit in France this year.
“The AI future is not going to be won by hand-wringing about safety,” he said.
Opinion: Is Canada’s AI Minister just an industry spokesman now?
On Wednesday, Prof. Hinton joined his friend and former University of Toronto student Nick Frosst, who is a co-founder of Cohere Inc., for a discussion about AI at the school.
Prof. Hinton is a revered figure in the field and since retiring from Google in 2023 he has become outspoken about the serious risks he believes the technology poses, including misinformation, job losses, more advanced cyberattacks and even replacing humanity.
“In early 2023 I realized that the digital intelligences we’d already created, although they weren’t as smart as us yet, might actually be a much better form of intelligence than biological intelligence,” he said. (He did, on a positive note, highlight the potential for AI to drastically improve health care.)
Mr. Frosst pushed back against some of his former mentor’s assertions, including that AI models will enable bad actors to more easily build deadly biological weapons.
Mr. Frosst countered that information is not the primary roadblock to building a bioweapon. Instead, he said bad actors have not been able to be do so “because they don’t have access to a wet lab, because they can’t run experiments, because they can’t get the right materials.”
There is intense disagreement about the dangers posed by AI, when those threats could emerge, and what to do about it – even among the country’s top AI scientists.
Yoshua Bengio, a Turing Award winner and the scientific adviser at the Mila - Quebec Artificial Intelligence Institute, is also vocal about the risks of AI and a proponent of regulation. He recently co-founded LawZero, a non-profit that will develop safe AI systems.
Canada’s first AI Minister Evan Solomon on his plan to scale up the industry
Richard Sutton, another Turing Award winner and the chief scientific adviser at the Alberta Machine Intelligence Institute, has a very different view.
“It’s too soon to regulate this new technology. And there’s no real harms,” he said in an interview after the event at U of T on Wednesday.
“You should let people be free to explore and innovate and experiment, and regulate afterwards, if needed.”
Prof. Sutton also criticized the AI safety field, which aims to mitigate the perceived harms of the technology through technical measures and policies.
Last year, the federal government allotted $50-million over five years to create the Canadian Artificial Intelligence Safety Institute to further research.
“I just decry that,” said Prof. Sutton, who also teaches at the University of Alberta. “It’s supporting this meme that AI is unsafe.”
An excessive focus on safety and regulation is “really bad for the field,” he said. “Having to ask permission to do something is extremely suffocating, and it just makes everyone afraid.”