- By Zoe Kleinman
- Technology editor
If you don’t know where you stand on artificial intelligence, you’re not alone.
Will it save humanity or destroy us? The stakes are that high and the jury is still out, even among the world’s leading experts.
Some AI creators are calling for its development to slow down or even pause altogether because it’s evolving so quickly. Others argue that doing so will mean we cramp the tech’s potential to achieve amazing things, like generating formulas for new medicines and work on climate-change solutions.
This coming week, around 100 world leaders, tech bosses, academics and AI researchers are gathering at the UK’s Bletchley Park campus, once home to the codebreakers who helped secure victory during World War Two. Their purpose is to take part in discussions about how best to maximise the benefits of this powerful technology while minimising the risks.
The event is the UK’s AI Safety Summit, and the risks it intends to focus on are pretty extreme. They relate to so-called “frontier AI”, the most advanced and powerful systems, which don’t currently exist but AI is advancing so rapidly that they may do soon.
But critics say the focus of the two-day meeting should be on the more immediate and pressing problems of AI, such as the large amount of energy it consumes and the impact it is already having on jobs.
Perhaps surprisingly, given this nightmarish backdrop, Prime Minister Rishi Sunak urged people “not to lose sleep”. He has a plan, and it’s an ambitious one. He wants to position the UK as the global leader for AI safety.
When the summit was first announced, and billed as a world-first, there were quite a few raised eyebrows. Would the world’s top brass actually travel to a remote, leafy corner of England, in the winter cold and close to the US Thanksgiving holiday, just because the UK says so?
No official guest list has been published. But it’s pretty clear by now that the US tech giants will be very well represented. Not always at CEO level, but high-level execs nonetheless.
There’s no shortage of summit enthusiasm from the commercial sector. British-based Stability AI boss Emad Mostaque described the summit as a “once-in-a-generation opportunity” for the UK to unlock AI superpower status.
“We will encourage the government and other policymakers to commit to supporting AI safety right across the ecosystem, from corporate labs to everyday researchers and from long-term threats to short-term risks to keep Britain safe and competitive,” he gushed.
These firms are an essential part of the discussion – they are at the front of the AI race and they are the ones building the systems. But you might imagine these are conversations they are already having among themselves. Diversity of thought here is crucial.
The world-leader contingent is a bit more of a mixed bag. US Vice-President Kamala Harris is attending, Canadian PM Justin Trudeau is not. European Commission president Ursula von der Leyen will be there, German chancellor Olaf Scholz won’t. China has been invited – controversially so, given its difficult relationship with the west. But it is still undoubtedly a tech superpower.
United Nations Secretary General Antonio Guterres is also going – which is interesting because there are growing calls for a global body to take on AI oversight.
Some experts fear that the summit has got its priorities wrong. The risk of extreme doomsday scenarios are comparatively small, they argue, and there are more immediate threats, far closer to home, which will be more worrying for many people.
“We’re concerned about what’s going to happen to our jobs, what’s going to happen to our news, what’s going to happen to our ability to communicate with one another. We’re concerned about the threats to people, communities, and frankly, the planet,” says Prof Gina Neff, who runs an AI centre at the University of Cambridge.
A recent study claimed the computing infrastructure required to drive the AI sector alone could use as much energy as that of a country the size of the Netherlands in just four years’ time. This will not be discussed at the summit.
We also know AI is already causing disruption to jobs. A friend of mine worked in a small marketing company. There were five copywriters – now there’s one, whose job it is to check the copy generated by ChatGPT. The Department for Work and Pensions is using AI tools to speed up benefits claims.
And what about the data used to train these powerful systems? The details are kept under wraps by the commercial companies who own them. But arguably, AI is only as good as the data it understands. We know bias and discrimination has crept into both off-the-shelf and bespoke tools that are available now.
This will also not feature in summit discussions.
What will the outcome of this two-day pow-wow be? You might hope the chiefs will emerge clutching a signed “Bletchley Park Agreement” and the world will be saved from the perils of AI.
We’ll probably have to save that for the movie version. The government itself says the feat of getting these people in the same room at the same time to talk at all is a success in itself – especially if China does show up.
Professor Yoshua Bengio is well-known as one of the three founding “Godfathers” of AI. He was asked at a recent event hosted by Chatham House in London, what he would like to see from the summit.
He suggested a registration and licensing regime for frontier AI models, revokable if a system is deemed unsafe – but acknowledged that something like this would take longer than two days to get together.
“We’re going to need to start with small steps that can be implemented quickly,” he said.
“International treaties and agreements take a lot more time… but we should start small and not wait to have built a very complicated global governance system before we start doing things.”