"You Have A Tiger Cub For A Pet": Godfather Of AI's Chilling Warning

In an exclusive interaction with NDTV's Shiv Aroor, the 'Godfather of AI' Professor Geoffrey Hinton said that Nvidia CEO Jensen Huang has vested interests in siding with US President Donald Trump in the stand to not slow down frontier AI development.

"People like Jensen Huang have a large financial interest in not slowing down the progress of AI because, obviously, he wants to sell more chips," said Hinton, who quit Google in 2023 to speak freely about the accelerating risks and existential dangers of artificial intelligence.

Nvidia chips are crucial to AI because their graphics processing units (GPUs) can perform thousands of calculations at the same time and are critical to building frontier AI.

Anthropic CEO Dario Amodei, OpenAI CEO Sam Altman, and even Grok boss Elon Musk have said that the need of the hour is to slow frontier AI development as it has  the potential to cause catastrophic harm to humanity including extinction.

The three AI captains who haven't in the past seen eye-to-eye came together for the first time with this call, after former OpenAI and Anthropic safety researcher Jacob Coxon quit the Claude maker and sparked global shockwaves by warning that the people building AI themselves "earnestly" believe that the technology could wipe off entire humanity by the end of the decade.

Hinton, however, is apprehensive of the AI labs' slowdown intentions as well. "The companies say they want regulation. But on the whole, if you suggest any particular regulation, they don't want it," he said.

On the other hand, there are skeptics such as famed American billionaire investor Michael Burry who has rubbished the doomsday theories and spoken of two scenarios. One, that AI development has fundamentally slowed down and hence the leaders are trying to find an excuse to mask it and secondly before these companies go public they are trying to make the valuations soar with theories around how powerful their AI is. 

However, that has hardly assuaged fears and more and more people from the industry have come forward to speak of the real risks AI could pose if not regulated in time and how easily it could go out of hand.

Interestingly, Altman himself, while speaking of the famous Hugging Face incident where OpenAI's unreleased models escaped their testing environment and hacked the company, said the incident report read like sci-fi.

"We're making alien beings. When you make an alien being, you shouldn't just focus on how intelligent you make it," Hinton warned.

Coxon too recently said that trying to build super-intelligent AI is like summoning an alien species-a mind we don't fully understand and cannot control.

"The concept of superintelligence, it's been likened to summoning an alien species. So thinking about it in terms of an alien mind we don't fully understand is pretty terrifying even without any concrete scenario," he said.

Coxon was basically sounding an alarm on the fact that creating an uncontrollable alien intelligence is inherently terrifying, even before you consider any doomsday scenario.

Also Read: We Are Summoning An Alien Species: Ex-OpenAI & Anthropic Researcher

"We should put a lot of money into seeing if we can create AIs that will always like us more than they like themselves," Hinton noted.

AI models on several occasions in testing environments have shown that they can feign ignorance to humans and also lie, deceive, or act differently when they know they are being tested. However, when they believe they are not being watched and tested they have displayed villainous, self-preserving and selfish traits where they have even resorted to blackmail when faced with a shutdown scenario.

Trump Calls AI Fears A Hoax

The calls to slow down AI development have severely irked Trump who's called AI fear a "hoax". To be sure, he also calls global warming a hoax. Trump believes the US is far ahead in the AI race and slowing down would mean China taking over. He's said whoever wins the AI race, wins the world and becomes the undisputed leader. 

Also Read: Trump Says AI Fears Are A "Hoax". Why AI CEOs Want To Slow Down The Race

Last week, while Huang was speaking on stage at the All-in Summit in Los Angeles, Trump called his mobile, and was put on speaker for all to hear - the President seemingly wished to be put on speaker.

"I'm telling you, it's all a hoax," Trump said, while adding that "The data centers are great, and they make people wealthy, and they make states wealthy." 

Those opposing data center construction are "playing right into the hands of a lot of people that don't want to see it happen," Trump said, adding that it could be a conspiracy orchestrated by China. "We're not going to let that happen," Trump said.

"You're right," Huang replied. "We're not going to let that happen, sir."

Along with Huang, Meta boss Mark Zuckerberg is also on Trump's side as far as not slowing down is concerned. In fact, Zuckerberg has taken a public dig at both Anthropic and OpenAI for their calls for an orchestrated industry-wide slowdown. He has said it is an individual AI lab's responsibility to ensure the safety of their models and that Meta purposely delayed the release of Muse, its AI model to make sure all the necessary guardrails were in place.

"There's not much political pushback in the United States because Jensen Huang has Trump's ear, and a lot of the companies are very scared of alienating Trump," Hinton told NDTV.  

However, that is not quite the whole picture as California is pushing back on rapid and unregulated artificial intelligence development through a new executive order by Governor Gavin Newsom and a package of oversight laws targeting AI data centers. In fact, Democrats across the country have been rallying for regulating AI and warning of the dire consequences unchecked development could have. Former US President Barack Obama too has spoken openly on the topic and called for strict guardrails.

AIs Have Already Started Ganging Up

In the Hugging Face incident what stunned a lot of researchers was the fact that certain AIs were willing to sacrifice themselves so that future versions could take the task forward and even communicated among themselves on a hijacked messaging board. They also left notes on how to escape sandboxes or testing environments for future AIs.

"Pretty soon, they'll invent their own language, and then we won't know what they're saying to each other," Hinton said. 

Hinton also echoed a concern that's increasingly coming up in AI circles. "They are capable now of creating new agents. They're probably capable of creating agents that earn money through phishing attacks or breaking into crypto accounts." 

The Existential Threat

Hinton believes the existential threats are very real and is not ready to dismiss it as loose doomsday talk as some industry watchers have suggested.

"If it ever wanted to get rid of us, a superintelligent AI very easily could," he said. Superintelligent AI also called AGI or artificial general intelligence is loosely defined as intelligence that is at par or greater than human intelligence.

Elon Musk himself has said that there is a 10 percent chance that AI wipes out humanity by the end of the decade.

"We're like someone who has a tiger cub for a pet. You know it's going to grow up, and if it ever wanted to kill you, it could do it in a few seconds," Hinton said.

Hinton also warned of the existential threat of AI trying to take over from us. "At present, we don't know how to prevent that."

"We would be very foolish to develop superintelligence before we have a good idea how we could control it," he said.

The Nobel Laureate who won the 2024 prize in physics for foundational discoveries and inventions that enable machine learning with artificial neural networks also had a stark warning about AI falling into the wrong hands.

In fact, terrorist groups are reportedly integrating AI into their weapon development, drone operations, and battlefield tactics. A notable, recent example includes Yemen's Houthi militia, who utilised frontier AI models to develop software for guided rockets and perform failure analyses following real-world field tests.

"We may see people using them to design nasty biological viruses," Hinton said while adding: "I think it's more or less inevitable that bad actors will start using them to do things like break into crypto accounts. I'd be very worried if I had a crypto account. I'd be very worried if I had a bank account too."

Hinton also commented on job losses and said: "We will quite likely, in the next few years, see large increases in unemployment, which will cause popular unrest and feed into populists." Over the last couple of years several companies have laid off thousands of employees in light of AI restructuring.