What do we Know About Labour’s AI Policy, and will it Work?

SUMMARY

• AI policy is central to Labour’s agenda at all levels: their search for economic growth; modernising the British state; and achieving ambitious efficiency plans.
• Labour’s AI Opportunities Action Plan commits to increasing the UK’s computational capacity by twenty times; AI Growth Zones; a National Data Library; and a focus on AI/digital skills.
• But there are practical challenges: AI requires extensive hardware and energy infrastructure which the UK lacks; AI requires data to train and refine models, but this poses infrastructural and also highly contentious political obstacles; and AI may entrench existing spatial inequalities.
• In sum, there is reason to be sceptical of Labour’s techno-solutionist approach.

Since taking office, the Starmer government has been bequeathed myriad political problems, compounded by the lack of a joined-up vision to address them. In this context, Labour appears to have latched on to the promise of artificial intelligence (AI). AI is increasingly central to Labour’s ‘rewiring the state’ agenda as well as its search for economic growth. In this sense, Labour’s policymaking is marked by a techno-solutionism, which casts all manner of political problems as amenable to computable solutions. Actually delivering AI-oriented political and economic solutions, however, is itself riddled with many contradictions and obstacles. As such, there is good reason to be sceptical of the potential of Labour’s techno-solutionist approach.

What is Labour’s AI agenda?

Labour’s AI Opportunities Action Plan sets out a crosscutting agenda to enable the proliferation of AI tools across the economy, society and the state. Labour have committed to increasing the UK’s computational capacity by twenty times; identifying a number of AI Growth Zones which will enjoy streamlined planning procedures and improved access to energy infrastructure; launching a National Data Library to boost dataset availability; and have redoubled efforts to improve the UK’s AI/digital skills mix.

Alongside this, the government has reconstituted the Government Digital Service as a one-stop hub for AI implementation in the public sector, begun piloting the Humphrey AI suite within the Civil Service, and has bet heavily on the identification of AI-driven efficiency savings in their latest Spending Review. AI is therefore central to Labour’s agenda at all levels: it is central to the ‘number-one mission’ of driving economic growth; is potentially key to rewiring and modernising the British state; and is instrumental in achieving an ambitious set of efficiency plans to ensure government spending remains within the bounds of the government’s fiscal rules.

That being said, the current government is not the first to emphasise the growing importance of AI. Successive British governments have hailed the transformative potential of AI for almost a decade. The May government’s Industrial Strategy can be considered the starting point for the UK’s AI obsession. It considered AI and the data economy as one of the four grand challenges facing Britain, and launched an AI sector deal to grow the UK’s AI sector. Though AI was less central to the nebulous levelling-up agenda which superseded the industrial strategy, the Johnson government doubled down on commitments to grow AI in the UK through the launch of its National AI Strategy. Under Sunak, the UK hosted the first global AI summit and focussed heavily on boosting UK AI R&D capacity and AI regulation in order to establish a distinctive regulatory approach distinct from both the US and the EU.

The challenges facing AI in the UK

Yet, successive governments have also struggled to implement their AI agendas and overcome political and economic obstacles that stand in the way of achieving these potentialities. Instead, UK AI has been dogged by years of policy hyperactivity. Highly ambitious agendas have been launched, but have been short-term, under-evaluated, reactive, fragmented, incremental, and top-down. This has resulted in policy failure and a high level of policy churn which has left key underlying issues unaddressed, resulting in a gradual erosion of the UK’s competitive advantage on AI. Three key areas are worth highlighting.

1. AI requires extensive hardware and energy infrastructure – which the UK lacks

AI requires extensive computational hardware and energy infrastructure. The UK’s computational capacity in relation to AI lags behind world-leaders like the US and China. Only three percent of global computational capacity resides in the UK, and much of this is not optimised for AI. Furthermore, electricity to power this infrastructure is more expensive in the UK than in the US, East Asia and Europe and the UK’s aging national grid is replete with storage and connection issues. Plans to turbocharge AI would therefore require huge investment in infrastructure after years of neglect. Yet, even if AI-specific infrastructure is delivered, the knock-on effects on other ailing infrastructure like the national grid represent further bottlenecks.

2. AI requires data to train and refine models

Second, AI requires data in order to train and refine models. Yet, most public sector data in the UK remains locked behind out-of-date legacy IT systems. Beyond this, the question of data availability raises difficult questions related to intellectual property. Various high-profile British creatives have called on government to ensure intellectual and creative property is protected from AI models in the face of attempts by the government to loosen these protections in the name of boosting AI. Thus the path to greater data availability poses further infrastructural and also highly contentious political obstacles.

3. AI may entrench existing spatial inequalities

Third, there is the issue of diffusion. Many of the loftier possibilities associated with AI rely on its diffusion throughout the economy to unlock productivity gains. On this front, the UK has struggled. Smaller firms lag behind on AI adoption and there are wide sectoral disparities between the already most productive sectors (IT, legal, financial services) where AI adoption is relatively high, and the already least productive sectors (hospitality, health, retail) where it is low. Both dedicated AI firms and the sectors where AI adoption is highest are concentrated in London and the South-East. This means that patterns of AI-driven growth and productivity gains may entrench existing spatial inequalities, rather than overcome them.

Will Labour’s AI agenda work?

In sum, the current government’s techno-solutionism elides myriad practical challenges related to infrastructure, data and diffusion. These issues become even thornier when one considers the British state’s inability to deliver major infrastructure projects on time or within budget. The pressing need to decarbonise and achieve net zero also looms large here, given the intensive energy demands of AI. So too do the potential knock-on effects of AI on ever more salient issues such as civil liberties and regional inequality. In this context, the imaginary of an AI-driven hyper-productive economy and hyper-efficient state seems far removed from the litany of political issues which are sure to emerge as the pandora’s box of AI is opened by a government in search of solutions to its political challenges.