close
close

Mondor Festival

News with a Local Lens

Fake content is getting harder to detect, but Hinton has an idea to make it easier
minsta

Fake content is getting harder to detect, but Hinton has an idea to make it easier

TORONTO — Artificial intelligence pioneer Geoffrey Hinton says it’s increasingly difficult to distinguish technology-generated videos, voices and images from the real thing, but he has an idea to help in the battle .

This increased fight has contributed to a shift in how the British-Canadian computer scientist and recent Nobel laureate thinks the world could combat fake content.

“For a while, I thought maybe we could call things AI-generated,” Hinton said Monday at the inaugural Hinton Conferences.

“I think it’s more plausible now to be able to recognize that things are real by taking a code there and going to certain websites and seeing the same things on that website.”

He believes this approach would help verify that content isn’t fake and imagines it could be particularly useful when it comes to political video ads.

“You could have something like a QR code (taking you) to a website, and if there is an identical video on that website, all you have to do is know that website is real” , explained Hinton.

Most Canadians have spotted deepfakes online and nearly a quarter encounter them every week, according to an April survey of 2,501 Canadians by Dais, a public policy arm of the University metropolitan Toronto.

Deepfakes are digitally manipulated images or videos depicting scenes that did not occur. Recent deepfakes have depicted Pope Francis in a Balenciaga puffer jacket and pop star Taylor Swift in sexually explicit poses.

The Hinton Lectures are a two-evening event that the Global Risk Institute is hosting this week at the John WH Bassett Theater in Toronto.

On the first night, Hinton, often called the Godfather of AI, took the stage briefly to remind the audience of the litany of risks he has warned the public over the past few years that the technology poses. He believes that AI could cause or contribute to accidental disasters, unemployment, cybercrime, discrimination, and biological and existential threats.

However, most of the evening was devoted to a lecture by Jacob Steinhardt, assistant professor of electrical engineering, computer science and statistics at UC Berkeley in California.

Steinhardt told the audience that he believes AI will advance even faster than many think, but that there will be surprises along the way.

By 2030, he imagines that AI will be “superhuman” in mathematics, programming and hacking.

He also believes that large linguistic models, which underpin AI systems, could become capable of persuasion or manipulation.

“There is significant room for maneuver if someone were to try to train them in persuasion, perhaps either an unscrupulous corporation or a government keen to persuade its citizens,” Steinhardt said. “There are many things you could do.”

He told the audience that he considers himself a “concerned optimist,” who believes there is a 10 percent chance that technology will lead to the extinction of humanity and a 50 percent chance that it will generates immense economic value and “radical prosperity”.

Asked at a later press conference about Steinhardt’s “worried optimist” label, Hinton called himself a “worried pessimist.”

“Research shows that if you ask people to estimate risks, normal, healthy people vastly underestimate the risks of really bad things…and the people who understand the risks correctly are the mildly depressed ones,” said Hinton.

“I consider myself one of those, and I think the risks are a little higher than what Jacob (Steinhardt) thinks, let’s say about 20 percent.”

Hinton also used the press conference to share more about what he and the Princeton University researcher did with his half of the 11 million Swedish crowns (about $1.45 million Canadian) , John Hopfield, received when they won the Nobel Prize in Physics earlier this month.

Hinton said he donated half of his share of the prize to Water First, an organization in Creemore, Ont., that trains Indigenous communities to develop and provide access to clean water systems.

He first considered donating some of the money to a water organization that actor Matt Damon is involved with in Africa, but then said his partner asked him, “What about Canada?” »

This led Hinton to discover Water First. He said he was compelled to donate because of the land acknowledgments he hears at the start of many events.

“I think it’s great that they recognize (who lived on this land first), but it doesn’t stop indigenous children from having diarrhea,” he said.

Hinton previously said a portion of his winnings would also be donated to an organization that provides jobs for neurodiverse young adults.

This report by The Canadian Press was first published October 28, 2024.

Tara Deschamps, The Canadian Press