Starmer’s AI revolution

By Morgan Watkins

Will the Prime Minister’s decision to make the UK an “AI Superpower” bring about the change he and his government have been promising? In May of last year, Labour announced their campaign slogan for the election ahead: “Change.” One word which aimed to make a nation, tired of recent political events, feel seen. With the economy only producing a meagre uptick of 0.1% last November, and the chancellor Rachel Reeves coming under strict scrutiny in recent weeks, Starmer is clearly feeling the need to prove himself and his government. Is the Prime Minister’s decision to head up the AI fan club really bring about the change he needs, and what dangers could this pose?

Firstly, there is the cost. AI computing power does not come cheap and considering how clear the government have been since the Autumn Budget regarding how little wiggle room there is in public finances, any extra funds devoted to AI will be fiercely criticised by those who would wish the money went elsewhere. In August, Labour cancelled a similar sounding £1.3 Billion plan to construct more publicly owned super computers in Edinburgh, Cambridge and Bristol, due to how it was “unfunded”. Labour’s new plan has committed to “increase compute capacity by twenty-fold by 2030” but is devoid of any exact figures.

AI computing power does not come cheap and considering how clear the government have been since the Autumn Budget regarding how little wiggle room there is in public finances, any extra funds devoted to AI will be fiercely criticised by those who would wish the money went elsewhere

The main funding streams we know about are solely coming from private investment, fast- tracked by government, to create “AI growth zones”. However, there is concern that as a country, we may not have the infrastructure required to support such a scheme. Such zones, as currently planned, will take up huge amounts of power, more than we currently have capacity for. The technology secretary, Peter Kyle, told The Times that one zone would be “building up within probably two to three years to 500 megawatts,” enough for 285,000 households. To go along with this plan, the Prime Minister announced a new AI Energy Council to design energy solutions such as small nuclear reactors, (SMRs) although none are currently built in the UK and designs are yet to be approved.

Beyond infrastructure, the government’s plan has ethical issues. Productive AI requires data sets on which it is are trained. As such, the government’s ownership of huge swathes of data sets, much of which containing (anonymised) data about the public, could prove to be a
massive boon to tech companies. Current plans include packaging data from the National Archives, British Library, and Natural History Museum which could be sold to companies as training data for new AI models. Although these would be uncopyrighted works, there are
potential issues around letting AI companies profit from the work of uncredited individuals and who never imagined their work would be used for this purpose.

Sir Keir is also keen to prove that AI may solve many solve current issues regarding public health. It is hoped that by adding the NHS’s anonymised health records to this AI data set, it could be used develop AI models which would assist in diagnosing patients. The Prime
Minister has emphasised that AI would only be used in such a way that would free up medical professionals from menial administration tasks, to allow them to gain more interaction with patients. However, would the public’s faith in the NHS waver if they knew that part of their diagnosis was coming from an opaque AI? Furthermore, in times when the Treasury will inevitably have to make cuts, would it be such a stretch to predict that AI may be used in such a way that results in a patient’s treatment being decided on by more artificial intelligence than human intelligence?

Unlike the previous government’s emphasis on AI regulation and safety, the PM has said that, we should “test and understand AI before we regulate it,” which suggests a ‘implement first, deal with unintended consequences later’ approach. In order for AI to be the magic pill labour clearly wants it to be, it must be used in a way that mitigates worries of privacy and in a way that creates more jobs than it makes redundant, which, as it stands, is a hard position to prove.

Image credit: Wikimedia Commons / Department for Science, Innovation & Technology