AI could fix our potholes – but also do Britain untold damage

In his readiness to be a ‘Goldilocks country’ for artificial intelligence companies, Keir Starmer has ensured that Britain will be at the forefront of the coming revolution – but we need to be careful not to roll over and allow tech investors free rein, says Chris Stokel-Walker

So, Britain is going “all-in” on artificial intelligence, then. In unveiling its “AI Opportunities Plan”, Keir Starmer has signalled his government’s intention to be a global leader in investigating and adopting the transformative potential of the tool.

The prime minister believes artificial intelligence has the potential to reshape the economy, and could enable his government to deliver on its mission to build an NHS fit for the future, usher in a “golden age of public service reform” and a new era of scientific breakthroughs.

In his enthusiasm, he has echoed his Labour forebear, Harold Wilson, whose head was similarly turned by the “white heat of technology”. When Wilson announced, some 60 years ago, that his administration would lean in to the coming “scientific revolution”, it was in order to modernise British industry and drive economic progress.

In its readiness to adopt, deploy and integrate AI into government business, Starmer has signalled that his administration’s involvement will ensure Britain is at the forefront of shaping technical standards internationally and its ethical development.

But, despite the government claiming it is “agreeing to take forward all 50 recommendations” set out in the long-awaited report by its tech adviser, Matt Clifford, the reality is not quite as clear cut.

Government press departments, as government press departments are wont to do, has been clever in its deployment of weasel wording. One recommendation – to relax visa rules to allow high-skilled AI experts easier access to the country – has neither been accepted nor rejected, and only “partially agreed”.

Nevertheless, this is perhaps the most enthusiastic welcome for AI from any government. And for good reason: the opportunities that can ensue from canny adoption of the technology will be significant. Starmer has talked about the way AI can help slash waiting lists, streamline planning applications, and even smooth the country’s roads by pinpointing potholes in need of repair. We’re going gung-ho on AI, the way forward charted out by Clifford, who is himself a big backer of the tech.

However, Clifford is a venture capitalist, and the risk is that the route we’re following under his guidance is one that will help big business as much or more than it will help individuals. In an op-ed for the Financial Times, Starmer asked those working at the frontier of AI to “take a look at Britain”, touting it as a place to invest, a place to build, and a place to bring your business.

The issue is that by laying out the red carpet for business, we could be creating an environment that runs roughshod over pre-existing rights and laws. The UK government is already consulting on changes to the copyright regime in the country that would materially weaken the rights of creative individuals in the face of big tech’s voracious appetite for “training data” for its systems. Barely changed carbon copies of creative works could soon be churned out by AI systems in the UK, if the government’s plans go ahead, all with impunity and the overt agreement government support provides.

There are also potential dangers in full-bore adoption of AI in other areas of our lives. Much has been made in the government’s announcement of the opportunities AI could bring to make the energy system more efficient – without much acknowledgement that embracing AI also itself has an environmental and energy impact, thanks to the huge amounts of power it takes to churn through big data.

The NHS was also seen as an area of improvement for AI, which is true – but it runs the risk of bringing about a two-tier healthcare system where if you pay, you see a human being, and if you don’t, you’ll be triaged and treated by a chatbot.

Also inherent in AI systems are the risks of biases – a worry, especially if AI is being considered for rollout across government services, as Starmer has indicated today. No one would have qualms with a mildly malfunctioning pothole detection algorithm. Plenty of people would if it’s used to decide who gets benefits, housing, or other support from councils or government departments.

This is where the problem lies. The prime minister has indicated he wants to be a “Goldilocks country” for AI companies – less stringent than the European Union, more regulated than the United States. The issue is that the approach he has taken has been developed by an ultimate tech insider, and it is nakedly open in trying to attract investment first, and considering the impact on the population second.

Chris Stokel-Walker is author of ‘How AI Ate the World: A Brief History of Artificial Intelligence – And Its Long Future’ (£14.99, Canbury Press)