AI pioneer Yoshua Bengio wants to focus on safer use of AI, thinks regulation is necessary

TORONTO – Artificial intelligence pioneer Yoshua Bengio says he will redirect his research to ensure he works on applications of the technology that are safe for society.

“I’ve been thinking about this a lot and I’m going to do what I think is best to move in the right direction,” the 2018 AM Turing Award winner and director of the Mila AI Institute in Quebec said Thursday night.

“I think I’m going to refocus my research so that I’m either working on applications that aren’t dangerous or very safe, like…healthcare or the environment, or on security, to prepare for and prevent what might happen. “

Bengio’s comments came at the end of a Munk debate hosted at Roy Thomson Hall in Toronto, where he and Massachusetts Institute of Technology physics professor Max Tegmark sparred with other technology experts over whether AI poses an existential threat.

The pair argued that AI is on track to become even more powerful and could achieve goals that are not in line with humanity’s best interests.

If someone with malicious intent gets their hands on the technology and the proper safeguards are lacking, the outcome could be catastrophic, they said.

“The amount of damage it can do to our society is also growing exponentially,” Tegmark said.

He pointed out that AI can spend years researching or reading the entire internet for hours on end and can be purposeful, meaning it can persuade, manipulate and build.

“For example, it could explore how to create more powerful bioweapons or how to create even more intelligent systems so that it can better itself personally,” Tegmark said.

“So naturally it would have the power and potential to wipe us out and achieve those goals.”

As part of his arguments, Bengio said humanity needs to take care of all the possible drawbacks of AI and move towards regulation because anyone using the technology with malicious intent could face “terrible consequences”.

“We have weaknesses. There are many ways we can dilute and there are conspiracy theories and many people believe them,” said Bengio.

“There will be people behaving strangely and doing things that can be very harmful.”

However, fellow AI pioneer Yann LeCun and Davis Professor of Complexity at the Santa Fe Institute Melanie Mitchell countered that AI can be used for good and lacks the authority to make the kinds of decisions that would pose an existential threat.

LeCun, who is also vice president and chief AI scientist at Facebook and Instagram parent company Meta, even said AI will bring “a renaissance” and “enlightenment.”

“Bad guys can use AI for bad things, but there are a lot more good guys out there who can use the same, more powerful AI,” he argued.

Mitchell added that AI learns from human data but still lacks fundamental aspects “of what it is to understand the world”.

“It’s a fallacy to think that a machine could be ‘smarter than humans in every way,’ and still lack common sense to understand humans,” she said.

“You would never give uncontrolled autonomy and resources to an AI that lacks these basic aspects of intelligence. It just doesn’t make any sense.”

In recent months, LeCun, who won the AM Turing Award alongside Bengio and AI pioneer Geoffrey Hinton, has had a different view of the risks of AI compared to his fellow award winners.

LeCun has said AI will surpass human intelligence, but doesn’t believe it will happen any time soon.

Meanwhile, in March, Bengio joined more than 1,000 technology experts in calling for a six-month pause in training AI systems more powerful than GPT-4 — the grand language model behind San Francisco-based OpenAI’s ChatGPT.

Signatories included engineers from Amazon, Google, Meta and Microsoft, as well as Apple co-founder Steve Wozniak and Rachel Bronson, president of the Bulletin of the Atomic Scientists.

Amid calls for a careful approach to AI, Ottawa is moving toward passing a law on artificial intelligence and data.

The law, part of a broader consumer privacy and data protection bill, would ban “reckless and malicious” use of AI, establish oversight by a commissioner and the industry minister, and impose financial penalties.

Ottawa has said it will go into effect no earlier than 2025.

This report from The Canadian Press was first published on June 22, 2023.

PART: