Elon Musk has revealed he expects artificial intelligence (AI) to be used by governments across the world to develop weapons before anything else.
The owner of Tesla, Space X and Twitter was speaking via video link at the annual Wall Street Journal’s (WSJ) CEO Council Summit in London on Tuesday.
The event, held annually across two days, sees chief executives speak about how they navigate the world of business in front of a global audience.
Opening the event, in an interview with the WSJ’s Thorold Barker Mr Musk said: “I think we generally operate with the much of an assumption that civilisation is robust and nothing can be brought down – a sentiment that has been common throughout history among empires shortly before they crumble.
When asked if AI advances the end of an empire, he replied: “I think it does. I don’t think (AI) is necessary for anything that we’re doing.
“There is a risk that as AI either improves it, either eliminates or constrains humanity’s growth.”
He continued: “There is an element of super-intelligence is very much a double-edged sword. If you have a genie that can grant you anything that presents a danger.
“I suspect the first government uses of AI to be weapons. So just having more advanced weapons on the battlefield that can react faster than any human could is really what AI is capable of.
“Any future wars between advanced countries or at least countries with drone capability will be very much the drone wars.”
Earlier this month the owner of Twitter announced he would be stepping back as chief executive of the company and revealed Linda Yaccarino would stepping in to focus on the operations of the business while Mr Musk remained in charge of product design and new technology.
He has now justified his decision to introduce a subscriber model where people can optionally pay for an enhanced service called Twitter Blue, with the 51-year-old saying it is “10,000 times” more difficult for AI to create accounts through a subscriber model.
Mr Musk said: “One of the first places you need to be careful of where AI is used is social media to manipulate public opinion.
“The reason why Twitter is going to a primarily subscriber-based system is because it’s dramatically harder, quote 10,000 times harder, to create an account that has a verified phone number from a credible carrier, that has a credit card and pays us a small amount of money a month and have those credit card numbers and phone numbers highly distributed and not clustered.
“It’s incredibly difficult. Whereas in the past someone could make a million accounts for a penny a piece and then manipulate or have something appear to be very much liked by the public when it is not, or promoted and retweeted when in fact the popularity is not there.
“The bias towards a subscription-based verification is very powerful and you won’t be able to trust any social media company that does not do this because it will simply be overrun by bots by such an extreme degree.”
When asked how big a shift it will be in the upcoming US election, he replied: “I think it’s something we need to be on the lookout for in the way of minimising the impact of AI manipulation.
“We’re certainly taking that seriously at Twitter and I think we’re putting in all the protections to detect large-scale manipulation of the system.”