The UK is leading the first global effort seeking consensus on how artificial intelligence (AI) might be regulated while attempting to identify risks and principles to ensure democracies are protected from the rapidly emerging technology.
The initiative led by prime minister Rishi Sunak will culminate in an international “AI safety summit” over two days from November 1st, and try to come to shared understanding whereby states recognise pitfalls and how they can be managed, while pursuing the benefits of AI.
This year, generative AI – notably the content creating ChatGPT – has provoked much fear that it is about to replace human thinking, besides undermining education, jobs and fuelling circulation of fake content.
The UK has been endeavouring to illustrate how AI models hold enormous potential to power economic growth, drive scientific progress and wider public benefits, while posing risks if not developed responsibly. Its effort is complicated by awareness the technology is going to drive the next digital transition, yet it is impossible to get an agreed definition of what it is.
“AI is already changing how we live and lives and we expect that to accelerate in the next year or two. There are big opportunities, but also considerable risks,” said British ambassador to Ireland Paul Johnston after a briefing in Dublin.
Government ministers, business leaders, researchers and other experts are gathering for “the first truly global AI safety summit”, Mr Johnston. It is being supported by the Irish Government.
“To trace the links back to the origins of computing, it’s taking place at Bletchley Park, where geniuses like Alan Turing helped break the Nazi’s Enigma Code during the [second World] War,” Mr Johnston said.
“There will be a focus on advancing international collaboration to identify and mitigate risks from frontier AI, the leading edge of AI research that is still unexplored and considered to be the most challenging. There will be space to debate the impact that may have on our shared safety.”
In a speech on Thursday, Mr Sunak noted: “AI will bring new knowledge, new opportunities for economic growth, new advances in human capability, and the chance to solve problems we once thought beyond us. But it also brings new dangers and new fears.”
[ Starkest warnings about AI have come from those building it. We need to listenOpens in new window ]
Fears need to be addressed head on, he said. “Doing the right thing, not the easy thing, means being honest with people about the risks from these technologies,” Mr Sunak said.
The UK issued a report detailing current risks of AI including societal harms, misuse and “loss of control”. By 2025, it is likely AI will also help create “faster-paced, more effective and larger scale” cyberattacks, its intelligence services warned.
The summit will focus on AI misuse by non-state actors and risks where systems may act autonomously in a way that does not align with human intentions or values. It will also showcase where AI is already bringing benefits.
The summit will include discussions on risks posed by the integration of frontier AI on society such as election disruption, bias, crime and online safety, Prof Gina Neff, director of the Minderoo Centre for Technology and Democracy in Cambridge, which is advising the UK government, said.
Whether or not such systems threaten humanity is hotly debated. Another report by the UK government office for science concludes “many experts consider this a risk with very low likelihood and few plausible routes to being realised”. It says to pose such a threat, an AI would need some control over vital systems, such as weapons or financial systems.
Leading AI specialist Prof Barry O’Sullivan of University College Cork says AI safety is essential, but disagreed with “AI worriers” overstating the potential impact by comparing it to pandemics and nuclear war.
Prof O’Sullivan dismissed “irresponsible scaremongering” because the idea of “a superhuman AI” does not exist yet. The director of the SFI Centre for Research Training in AI told the Global Ireland Summit last week that humanity-level risks to be concerned about were set out under the United Nations sustainable development goals; ie climate change, global poverty and human rights.
Prof O’Sullivan has, however, acknowledged the potential military threat that could manifest itself by way of lethal autonomous weapons or deployment of “swarms of drones”, and cautioned AI had the potential to contribute to societal division if not applied appropriately.
Aside from the potential AI has for Irish tech, he said there was insufficient focus on AI and diplomacy.
“AI is impacting geopolitics every day, and impacting the practice of diplomacy,” which raised issues around deep fakes and undermining of democracy. “Ireland has a fantastic role to play in putting itself in the middle of these very complex discussions and being the honest broker, the country that is trusted... I don’t know of any other country is a position to do that,” Prof O’Sullivan said.
The UK summit is one of the ways to rebalance the conversation about AI and address trust issues, Prof Neff said at the Dublin briefing.
“We have the possibility of changing the focus to make sure we’re taking a reasonable approach to innovation and not shutting things down,” Prof Neff said.
“I’m very optimistic that there’s an enormous number of opportunities and change that we will see in our lifetimes from these powerful and transformative tools ... we have the opportunity to get them right.
“So I would say, let’s not fall victim to the fear mongering, but let’s be prepared and let’s be sure that we assert what’s important to us as a society.”
