AI is an ‘alien being’ and humanity is running out of time to adapt, says British-Canadian Nobel laureate

Article content

“It’s obvious that these AI models understand what they’re saying,” he told The Guardian, “You couldn’t possibly answer all the questions without understanding the question. It’s ridiculous to say they don’t understand.”

Article content

And while alien in one regard, Hinton also remarked to The Atlantic that AI is already “more like us than most people think” and that what has become known as AI hallucinations — when chatbots present false, misleading or entirely fabricated information as the factual truth — is more akin to “confabulation,” something humans are guilty of when trying to retrieve stored memories.

Article content

Some American Democrats have seized on the risks around AI as a big issue heading into the November midterms and the 2028 U.S. presidential election. Photo by KIRILL KUDRYAVTSEV /AFP via Getty Images

Article content

“If I ask you to remember something that happened a few years ago, you will reconstruct a memory that isn’t actually all that faithful to what actually happened. It’ll often have the main points correct, but many of the details of which you’re quite confident, you’ll be wrong about,” he said.

Article content

“But that’s how human memory is. We reconstruct things. We don’t extract them from some storage. Now, AI does exactly the same thing, and that’s why it confabulates.”

Article content

Article content

Article content

He’d also made a similar prediction during an appearance on British entrepreneur Steven Bartlett’s The Diary of a CEO podcast in July, noting he often goes as high as a 20 per cent “chance they’ll wipe us all out.”

Article content

In an interview with U.S. podcaster Ezra Klein shortly after, Nvidia CEO Jensen Huang dismissed the dire prediction, calling it “hurtful” and arguing that it wasn’t based on scientific research.

Article content

“Don’t think for a second just because you’re an alarmist that you’re doing a social good — it is not true,” he said.

Article content

Speaking to The Atlantic around the same time, Hinton said estimates like those are “a gut feeling in quantitative terms,” not something that can be easily estimated scientifically.

Article content

“So when, for example, someone says there’s a naught per cent probability of AI taking over, they’re being crazy. They’re saying it’s absolutely impossible this’ll happen. That’s just crazy,”  he said. “When people say there’s a 99 per cent probability it’ll take over, that’s just crazy, too. So it’s somewhere in between. But they’re to give you an idea this is not negligible.”

Article content

But should the ominous prediction prove true and the machines do take over, Hinton told The Guardian there is no do-over for humanity.

Article content

“We can’t say, ‘Well, we lost that time. Let’s rerun the game.’ It’s not like that.”

Article content

For his part, he hopes that whoever gets there first will develop a model that puts humanity’s interest above its own.

Article content

A trend throughout Hinton’s recent media engagements has been scathing criticism of U.S. President Donald Trump and the tech industry’s biggest players, which, like energy and tobacco companies before them, he said, can’t be trusted to police themselves.

Article content

During a fireside chat with Jeff Scully of the Jefferson Educational Society in Erie, Penn., this month, Hinton called Trump’s leadership in general “appalling.”

Article content

“He’s corrupt. He’s probably senile. He’s a convicted felon,” he said bluntly. “If he had to make a choice between the future of humanity and him and his family getting rich, which do you think he’d go for?”

Article content

Article content

Article content

Hinton reiterated some of those opinions to The Guardian, along with his insistence that Trump proved his lack of AI understanding by “officially” renaming it superintelligence or SI, a long-existing term used to describe yet-to-be-developed AI that is “smarter than people.”

Article content

Hinton and other AI experts met with U.S. Congressional leaders in September, offering the lawmakers a warning that they could have as little as a year to introduce protections before losing control of AI, NBC reported.

Article content

“If you look at predictions for when we’ll get superintelligence, it used to be maybe 30 years, maybe 50 years. Then it came down to maybe 10 years, maybe 20 years. Now people are saying, a lot of the researchers are saying only a few years,” he told reporters after the closed-door meeting.

Article content

Hinton later told The Atlantic that a starting-point suggestion to create something like a Food and Drug Administration for AI — an agency to ensure what’s released for public use is safe — “seemed to make an impression on the senators.”

Article content

Hinton, meanwhile, in his conversation with Scully, lauded Canada and Prime Minister Mark Carney’s resilience in the face of U.S. trade aggression and said he was hopeful for Canada’s recently announced National Council on Artificial Intelligence. 

Article content

Article content