The prime minister’s ambitious approach has many detractors, but he deserves credit for attempting global leadership in a key area for the UK, says Dr Matthew Gill
Next week, the UK will host an artificial intelligence (AI) safety summit at Bletchley Park. The run-up has been turbulent, with the summit receiving criticism for a narrow cast list and a doom-laden focus on hypothetical future risks. Gaining buy-in from world leaders has been difficult, with US vice president Kamala Harris confirming attendance only today and European Commission president Ursula von der Leyen still TBC. Yet far from distancing himself from the event, prime minister Rishi Sunak has doubled down with a major speech this morning,
Prime Minister's Office, 10 Downing Street and The Rt Hon Rishi Sunak MP, Prime Minister's speech on AI: 26 October 2023, 26 October 2023, www.gov.uk/government/speeches/prime-ministers-speech-on-ai-26-october-2023
alongside publication of a discussion paper on the capabilities and risks of “Frontier AI”.
Department for Science, Innovation and Technology, Frontier AI: capabilities and risks – discussion paper, 25 October 2023, www.gov.uk/government/publications/frontier-ai-capabilities-and-risks-discussion-paper
Attempting to lead the global response to AI is a bold move by Sunak – and a risky one.
Sunak wants to balance significant opportunities with significant risks
Sunak is a techno-optimist. He began his speech by setting out the opportunities AI presents, and the discussion paper claims that “the overarching risk is a loss of trust in and trustworthiness of this technology which would permanently deny us and future generations its transformative positive benefits”.
Nonetheless, the prime minister went on to list some eye-watering risks: that AI could make it easier to build chemical or biological weapons; to spread fear and destruction; or to facilitate cyber-attacks, disinformation, fraud, and child sexual abuse. While counselling that there is not yet cause for alarm, he also flagged the potential risk of a superintelligence emerging that could cause humanity’s extinction.
To address these risks, Sunak announced a UK-based AI Safety Institute to help the world evaluate new types of AI and understand the risks they may pose. He also proposed – for agreement at the summit – that the Intergovernmental Panel on Climate Change (IPCC) be used as a model for a global expert panel that could achieve international consensus on the state of AI science. But the prime minister also announced a package of measures to pursue potential opportunities, including a £2.5bn investment in computing power for use by businesses and researchers developing AI in the UK.
Sunak’s approach bears some hallmarks of good policy making, although there are gaps
The prime minister is – rightly – starting from a desire both that government should fully understand the facts before jumping to solutions, and that the strategic response to a technology that does not respect international borders should be internationally agreed. Sunak’s reluctance to draw conclusions too early about how to regulate AI may frustrate those worried by present harms, but it retains agility in the face of a fast-moving technology as well as the potential for global leadership by the UK.
The approach certainly misses some things – including, as the Institute will set out shortly, the need to build economic and social resilience in the face of change alongside a regulatory response to that change. It still leaves work to be done in reconciling Sunak’s techno-optimism with his concerns about extreme risks in practice. And arguably it seeks to run before it can walk: if the UK’s domestic position is not yet fully worked through, can it really lead the development of a global one? But the summit is creating a space for international discussion and collaboration which can be built on in future, and will also help the UK to prioritise developing its own regulatory response, building on a white paper that was well received earlier in the year. 12 Department for Science, Innovation and Technology and Office for Artificial Intelligence, AI regulation: a pro-innovation approach, 29 March 2023, www.gov.uk/government/publications/ai-regulation-a-pro-innovation-approach
But the summit is a bold and risky political move
There is a lot of uncertainty about the future of AI. There have been many over-hyped technological innovations in recent years. Sunak is right that government should attend to low-probability but high-impact risks – as Covid illustrated – but it will not help his legacy if, with hindsight, he is seen simply to have over-reacted. It is also unclear either that the UK is in a position to lead internationally in the medium term – with the US, EU and China all key players – or whether this prime minister in particular, as he nears the end of his current term, has the political stature to convene a global summit effectively.
Sunak might simply have concluded – with the UK facing many headwinds, and perhaps less than a year remaining in office – that trying to lead the global response to AI is his best chance at a lasting legacy as well as, perhaps, a welcome distraction from more prosaic problems at home. That could explain the rush to establish an international summit in advance of the UK having its national regulatory position clear. But Sunak should also be taken at his word, that he thinks this really matters, particularly since the summit is hardly likely to be a significant vote-winner.
What remains to be seen is whether a narrowly focussed and hastily organised global summit can significantly catalyse a co-ordinated global response to AI’s development. It is a gamble but if, in three years’ time, the UK finds itself at the forefront of an IPCC-style international network and secretariat setting the global agenda in this area, then Sunak will be able to look back on this summit as a gamble that paid off.