AI could 'kill many humans' within two years, Sunak adviser warns

Sign up for View email from Westminster for expert analytics straight to your inboxReceive our free email View from WestminsterPlease enter a valid email addressPlease enter a valid email addressI would like to receive emails about offers, events and updates day of The Independent. Read our privacy notice{{ #verifyErrors }}{{ message }}{{ /verifyErrors }}{{ ^verifyErrors }}An error has occurred. Please try again later{{ /verifyErrors }}

Artificial intelligence (AI) could create 'dystopia', a government minister has warned after Rishi Sunak's technology adviser said it could become powerful enough to "kill many humans" in just two years.

Matt Clifford said even the short-term risks were "pretty scary", with AI having the potential to create cyber weapons and biological weapons that could inflict many deaths. It comes as Mr Sunak travels to the US, where he is set to try to persuade President Joe Biden of his 'big plan' for the UK to be at the center of international regulation of the 'AI.

The Prime Minister wants Britain will host an AI watchdog similar to the International Atomic Energy Agency, and will also propose a new international research body.< /p>

Paul Scully, minister for technology and the digital economy, told the TechUK Tech Policy Leadership Conference on Tuesday there shouldn't just be a focus on a 'Terminator-style scenario'.

He said: "If we are wrong, there is a dystopian point of view that we can follow here. There is also a utopian point of view. Both can be possible.

"If you're only talking about the end of humanity because of a Terminator-style rogue storyline, you're going to miss how good AI is already working - how it's mapping proteins to help us with research medicine, how it's helping us fight climate change.

"All of these things it's already doing and will only get better.

"We need to take a little breathing space to make sure we're doing it right for society as a whole, as well as for the benefit of the industry." Clifford said that unless AI producers are regulated at scale world, then they could be "very powerful" systems that humans may find difficult to control.

Mr Clifford's intervention comes after a letter backed by dozens of top experts - including many AI pioneers - warning that the risks of technology should be treated with the same urgency as pandemics or nuclear war.

Senior Google officials DeepMind and Anthropic signed the letter with the so-called "godfather of AI", Geoffrey Hinton. Mr Hinton resigned from his role at Google earlier this month - warning that in the wrong hands, AI could mean the end of humanity.

Mr. Clifford advises Mr. Sunak on the development of a government task force that studies in AI language models such as ChatGPT and Google Bard, and is also president of the A...