Controversial tech entrepreneur Elon Musk told Rishi Sunak artificial intelligence (AI) was likely "the most destructive force in history" that mankind would witness and one day, "there will come a time where there are no jobs".
He spoke to the Prime Minister live on stage at an exclusive event at Lancaster House in Mayfair, London, before an audience of British tech startups, business leaders and X users, and celebrities including American rapper will.i.am.
Mr Musk, who has long warned of the dangers of AI, did caveat his comments by adding not all AI is bad: “On balance, I think AI will be a force for good mostly, but the probability of it going bad is not zero.”
The conversation came at the end of the Government’s two-day artificial intelligence (AI) safety summit, where 28 countries signed a declaration to collaborate on AI safety research on Wednesday, and a new plan was set out to test out AI models both before and after launch, for critical national security, safety and society harms.
The tech billionaire thanked Mr Sunak for including China in the AI safety conversation and said he had recently visited the country and had frank and serious conversations with Chinese lawmakers.
Mr Musk mentioned deep fake videos and disinformation from fake bot social media accounts, but in particular painted a horrifying sci-fi movie vision of humanoid robots, referencing James Cameron's The Terminator series.
"At least a car can't chase you up a tree, you can go hide indoors... but if you have a humanoid robot it can basically chase you anywhere. I think we should have some hardware, a local off switch. Any software that can be updated from the internet can be overridden," he said.
However he praised Mr Sunak for highlighting AI safety concerns: "I think even if there are no firm regulations, even if we have a capability to highlight concerns to the public, that would be very good. If that's the least that comes out of [the AI safety summit], that would be great."
The positives of AI
Mr Sunak asked whether open source software development would be preferable to private, proprietary software development, but Mr Musk said you might still not be able to work out what an open source AI was doing.
He said he feared that AI systems could still overwhelm society if not carefully monitored.
However he also spoke at length about the good aspects of AI, such as the ability for it to be used one day to give every child a "personal tutor" that could learn about and grow with them over time, or provide an AI companion.
"One of my sons has learning disabilities, has some trouble making friends. An AI friend would be great for him," he said.
"If you have an AI that has memory and remembers all your interactions and you give it permission to read anything you've ever done, so really it will know you better than you know yourself, you will actually have your friend. As long as that friend can stay your friend... but I think that will actually be a real thing."
Mr Musk, who owns electric car company Tesla, aerospace firm SpaceX and the social media platform X (formerly Twitter), praised London, stressing that it and San Francisco are "the two leading locations on Earth for AI".
He also praised British semiconductor firm ARM, now owned by Japan's Softbank, saying that it "is one of the best, if not the best in chip design".
"We use a lot of ARM tech, almost everyone does actually," he added.
Nevertheless, while he had a very magnanimous view towards AI in some regards, he was still worried the technology could easily overwhelm mankind without us being aware.
"The pace of AI is faster than any technology I've seen in history so far. It seems to be growing in capability by fivefold and tenfold per year... and Government isn't used to moving at that speed," Mr Musk warned.
On the future of X
Mr Musk said that he is worried about how AI will impact the future of social media platforms like X (formerly Twitter).
"The general aspiration with the X platform is to be as accurate as possible and as truthful as possible, even if people don't like the truth, and if we stay true to the truth, then people will use the system to find out what's going on and I think truth pays," he stressed.
"We might have to charge a very tiny amount of money to deal with the onslaught of AI-powered bots. That is a growing problem, perhaps it will be an insurmountable problem next year."
Mr Musk felt that "any social media" that doesn't taken action to address the issue of bots being used for disinformation "will be completely overrun by bots" in the near future.
For that reason he said it was prudent to possibly charge X users a £1 or $1 a year dissuade people from setting up armies of bot accounts to falsely make certain social media posts more prominent.
Mr Musk was asked by a member of the audience about whether it would be possible to develop technology that could "put a sense of trust" in media that goes viral.
Mr Musk agreed that such authentication could be useful, however he was vague about how that could be achieved. No media was allowed to ask questions at the event.
The tech billionaire also deflected a question from the audience asking how to prevent developing countries from being discriminated against if their pace of AI innovation falls far behind that of richer nations.
Mr Musk said he wanted to end the interview on a "positive note", so he emphasised the "utopian" nature of AI and the fact it would lead to higher average incomes across the population, as well as universal education.