As American and Chinese tech companies continue to push the frontiers of AI capability, it can sometimes feel utterly demoralising for an ordinary person looking on. If AI’s development worries you, if you have concerns about advanced AI systems doing harmful things beyond our control, or indeed have any anxieties surrounding the technology, you should know you’re not alone. Research by King’s College London indicates more negativity than positivity when it comes to public perceptions of AI in the UK, at least when it comes to jobs.
Along with this feeling of concern can come a deeply maddening sense of powerlessness, too. “What on Earth can I possibly do?” you may ask yourself after seeing yet another ‘oopsie’ moment from OpenAI or Anthropic wherein a model deceives its creators or commits acts of cybercrime. “How can I help?” As an ordinary person, the feeling is amplified by being outside of the two countries in which most AI progress is happening. The major firms are in China and the US, so for a person in the UK (or indeed any other country) that feeling that we’re helpless bystanders becomes ever more acute. Luckily, these feelings, though understandable, aren’t quite right.
Despite the fact that a tiny fraction of humanity does genuinely have enormously outsized influence over these consequential developments, ordinary people in Britain still have a few tools available to help push things in the right direction with regards to AI safety. The UK has already led the way on model evaluations. Our AI Security Institute (AISI) is ‘the closest the world has to an AI safety inspector’ according to one article in the Economist. It’s true that the institute has access deals with major American AI firms and has done vital research on things such as model political persuasiveness, user reliance on AI systems for emotional support, and they’ve supposedly even detected vulnerabilities in models that could have enabled their use as tools for making bioweapons. All of this is an excellent start, but it’s clear we need more. Much more.
As a middle power, the UK has the ear of both of the superpowers, at least to some extent. We have significant soft power through our language, arts, and R&D (although who can deny these have mostly declined in the last decade?), have one of the world’s largest economies, and still possess greater military capabilities than most countries. Very little of any of this has been used to usher in a safer path for AI. Yet, pressure is growing. Lord Tim Clement-Jones has proposed an AI Kill Switch amendment in the House of Lords, whilst Labour MP Alex Sobel and over 70 other MPs are asking the government to support their legislation to halt the development of Superintelligent AI in Britain.
This is all encouraging stuff, and it’s unlikely it would have happened without ordinary people in this country (and of course AI experts in organisations like PauseAI and ControlAI) applying pressure on politicians and raising awareness about the dangers of superintelligence. This should help reduce that sense of powerlessness we talked about. The more MPs and Lords are made aware of the issue, the more likely they’ll be to support a pause on frontier development and other AI regulations/protections. This, in turn, means more pressure on Downing Street to support such measures. Of course, legislating against ASI (Artificial Superintelligence) in the UK isn’t much of a winning strategy if the models are being built in the US and China, though it is a start. But to actually, meaningfully reduce the odds of a major AI-caused catastrophe, the UK must build international consensus for a global AI pause, not just a national one.
On this front, there is also precedent. The UK hosted the world’s first ever global summit solely focused on AI in 2023. Granted, the agreements were vague and non-binding, but the summit itself is a testament to how the UK can still be a force for good in the realm of global cooperation. There is precedent for countries coming together and curtailing the use or creation of dangerous technologies. The Biological Weapons Convention of 1972, the banning of chlorofluorocarbons in 1989, and the Nuclear Non-Proliferation Treaty of 1968 all point to the same conclusion: it is difficult, but not impossible, for geopolitical and economic rivals to agree to restrict their own technological advances when the risk to the world outweighs the benefit. There are counter-examples, of course, with the utter failure of the international community to deal with climate change being the most obvious of the bunch, yet the successes show an international pause on frontier AI development is at least possible. Indeed, Sobel’s bill includes ‘champion(ing) an international agreement’ on superintelligence.
The UK, with its close relationship to the EU, lukewarm relationship with the USA, and improving relations with China, is in a strong position to exert positive influence on the three major world powers in the direction of a pause. The EU has regulatory power, the US and China are the leading developers of the tech. All three need to work together (and with other nations too!) to create international oversight and rules on advanced AI development and deployment. The UK should host another summit, this time with the goal of developing an international treaty banning the creation of superintelligence. How this would be enforced (and other details) are beyond the scope of this article, but anyone interested can look at PauseAI’s policy proposals. The point is that Britain is an influential country and can use that influence to bring the world to the table and create opportunities for a pause and/or banning superintelligence. Our cultural soft power can be utilised too, generating greater public support for our cause and the values of human creativity. Our creative industry is still important globally; we should use it!
So, that feeling of powerlessness, has it gone away yet? I imagine not, and it might never fully dissipate, but the situation should be clearer now. The British government has the power to push the world in the right direction vis-à-vis AI safety, and that government is made up of MPs. Those MPs can be pressured by their constituents. You are one of those constituents. You are not powerless, and the UK is not powerless. Britain must make the case for an international treaty to pause frontier AI development, and ban the creation of superintelligent AI.