AI Safety Summit puts Rishi to the test

Since Brexit, Britain’s role as a major convener for international cooperation has been tarnished. This week, Rishi Sunak and the UK Government pitched to change this at the inaugural Artificial Intelligence Safety Summit, bringing together representatives from the USA, the EU, China and the U.N, to discuss international efforts to manage the risks of Artificial Intelligence (AI).

The event sought to combat the “catastrophic” risks of the technology, with Prime Minister Rishi Sunak stating that the potential harm posed by artificial intelligence were too significant to be left to big tech firms to consider alone.

The final day of the event is currently wrapping up at Bletchley Park, an English stately home symbolically chosen for being the site of the wartime code-breaking project in which Alan Turing’s work was instrumental. So, did Sunak’s summit pass the test?

Before zooming in on Bletchley, it’s worth considering why the UK Government has convened the summit. As AI innovation develops, which commentators believe is as important as the development of the printing press or the industrial revolution, many have highlighted the significant risks that the technology poses.

Leading experts believe that the uncontrolled and unregulated pace of innovation of the technology could present a risk to human life, with threats such as AI developing the ability to create bioweapons, the kinds of claims that would have been science-fiction only a decade ago. The summit was therefore convened to agree a set of measures to further the safety of global AI use, especially of so-called ‘frontier AI’.

The event brought together some of the leading minds in AI, with leading tech nations such as the USA and Japan, emerging tech powers such as India and Brazil, and leading minds from the tech community in attendance. Bringing together both USA Vice President Kamala Harris, and representatives from China, despite heightened US concerns about China’s threat to their national security, many see this as a diplomatic coup for the UK to get these representatives around the table, despite Joe Biden and Emmanuel Macron deciding not to attend.

A pact from the agreement, also known as ‘The Bletchley Declaration’, was one of the most significant outputs from the event, signed by 25 countries and the EU. This outlines a shared vision for building in safety and ethical considerations into AI developments and deployment, with focuses ranging from cooperation to ethics, to democratic values, and accountability. A glaring omission, however, is that a roadmap on integrating these principles was left out.

The more damning risks that the UK Government have focused on have prompted some to question whether they are barking up the wrong tree with their priorities. Why focus on bioweapons and potential algorithm-driven genocide tomorrow whilst job security risks are real today?

Critics have already argued that the declaration has been moulded by big tech companies. The CEO of machine learning startup Eigen Technologies, for example, has said that the apocalyptic warning “is overly influenced by a deeply flawed analysis and an agenda set by those big tech companies seeking to dominate the policy-making process. This kind of doom-mongering echoes the words of OpenAI and its peers, who have been among the most influential corporate lobbyists in the run-up to the Summit”, neglecting questions around IP, bias and governance are already critical considerations.

Overall, the summit has offered a moment of apparent cohesion in a seemingly increasingly fractured landscape of international cooperation, highlighting how governments around the world see the risks and opportunities of AI to be key priorities.

Rishi and his team will be pleased in achieving headlines around a ‘world-first’ AI agreement. Whilst there has been agreement, however, the real test will be its implementation. Critics’ snap reaction is that the UK Government has achieved a successful ‘photo opportunity’ rather than substantive debate on the most significant technology in generations.