Rishi Sunak has announced that the UK will host a global summit on safety in artificial intelligence in the autumn, as fears grow that the technology’s rapid advancement could spin out of control.
Safety concerns are mounting after breakthroughs in generative AI, which can produce convincing text, images and even voice on command, with tech executives such as Elon Musk among the figures expressing alarm. Here is a look at what the summit might achieve.
What is the aim of the summit?
The prime minister has changed his tone on AI in recent weeks. Having been overwhelmingly optimistic about the opportunities it creates, he has begun to talk about its “existential risks”.
Sunak is trying to position the UK as the natural hub for efforts to regulate the industry on a global scale, one that can provide a bridge between the US and China, and to offer an alternative to what some consider to be the EU’s heavy-handed approach.
Described as the “first major global summit on AI safety”, the government says it will consider the risks the technology poses and discuss how they can be mitigated through “internationally coordinated action”.
Is internationally coordinated action needed?
Industry professionals harbour concerns about AI and have issued warnings about the dangers it poses. Elon Musk was one of more than 50,000 signatories to a letter in March that called for an immediate pause in the development of “giant” AIs, alongside the creation of “robust AI governance systems”.
Concern about a possible existential threat from a system that human intervention cannot control by human intervention is not universal though. Many in the tech industry argue that the focus should be more immediate, for instance by focusing on the potential for generative AI, which can provide plausible imitations of text, images and voice that could produce destabilising disinformation during elections.
What would a global framework look like?
The UN-brokered treaty on the non-proliferation of nuclear weapons, which came into force in 1970, is an example of a global attempt to mitigate an existential threat. The treaty, to which 191 states are signatories, commits those that have nuclear weapon states to not helping those who do not acquire or build them. The International Atomic Energy Agency oversees compliance through inspections. The treaty also promotes the spread of peaceful uses of nuclear energy.
The letter calling for a six-month pause in AI development offers an insight into what might go into such a framework. It calls for dedicated regulatory authorities, public funding for safety research, and oversight and tracking of powerful systems.
Would a nuclear arms-style framework succeed with AI?
As with nuclear weapons, the technology that such a framework would seek to contain is already out there and proliferating. The chatbot phenomenon ChatGPT reached 100 million users within two months of its launch and a now-famous fake image of the pope wearing a Balenciaga jacket has underlined the power of generative AI to deceive.
One of Google’s engineers warned last month that the company could lose out to open-source AI technology. Such developers release their work for anyone to use, improve or adapt as they see fit, making it difficult for a framework to curb the use of open-source models.
after newsletter promotion
China is likely to be excluded from the UK summit, which will be open to technology companies and “like-minded countries”. This indicates that an international framework to place guardrails around AI development will not have the participation of a global tech and AI powerhouse.
What is happening in AI regulation already?
The EU has made significant strides in the area and is proposing legislation that will be seen a pacesetter for AI regulation, confirming the bloc’s status as a leading tech regulator.
The UK’s AI white paper sets out a set of principles to which the industry should adhere, but offers little in the way of concrete rules to govern it. Ministers disappointed some experts by resisting the idea of creating a new regulator to focus exclusively on AI.
The G7 has agreed to create an intergovernmental forum called the “Hiroshima AI process” to debate issues around fast-growing tools.
