Search site...

AI as threat or friend to workers? Rethinking the UK’s policy response through a capabilitarian approach to growth

Anastasia Siapka

26th March 2026

The threat of AI?

Advances in frontier technologies have sparked alarmist headlines about a workless future, urging UK policymakers to counteract automation. Matched with cost-cutting incentives, the growing sophistication and availability of Artificial Intelligence (AI) strengthen the threat of automation – particularly after COVID-19 and Generative AI. With work being a primary or main source of livelihood, this threat affects UK citizens horizontally but especially young people now entering the labour market.

Yet, framing AI-driven automation as a threat is neither accurate nor innocuous. It assumes that AI is a force capable only of disrupting society for the worse, of principal concern to policy as a potential cause of mass unemployment, overlooking the reverse possibility that AI could instead be ‘tamed’ to serve policy goals. It also undermines citizens’ agency. Treating AI’s adverse impact on work as inevitable obstructs workers from changing its course. For fear of being replaced with more profitable technology, they are more likely to accept worse wages and working conditions. Hence, the threat of automation – materialised or not – risks weakening workers’ bargaining power, exposing them to exploitation.

Pushing back against this framing, the UK needs a new policy agenda – one that, instead of fearfully avoiding or succumbing to AI, (pro)actively shapes and re-orients it towards positive directions.

Beyond economic growth

What should this agenda aim at? An intuitive answer is ‘growth’. Nonetheless, growth understood solely in monetary, material or productivity terms is incomplete: it disregards how wealth is distributed across citizens and how citizens convert wealth into well-being. Policies fixated on strictly economic growth, without considering such distribution and conversion variations, reinforce inequalities. I instead suggest a richer understanding of growth based on the Capability Approach (CA). Pioneered in international development, the CA is a normative theory increasingly applied to well-being and policy assessments. While the theory itself is well-established, its application to AI and the future of work remains unexplored.

The CA doesn’t ask how much wealth or other resources citizens possess but rather how these translate into what all citizens are really able to do and be. Its benchmarks are ‘functionings’ and ‘capabilities’. Functionings are what individuals value doing or being. Such doings and beings are simple (e.g., sustenance) or complex (e.g., self-respect). Although functionings capture actuality (what individuals are actually able to do or be), capabilities capture opportunity: they imply the functionings available to individuals. Hence, capability is ‘the substantive freedom to achieve alternative functioning combinations’.

In the CA, economic growth isn’t an end in itself but a means to a broader growth – that of human capability. While individuals are predisposed to develop their capabilities, they can only do so under favourable external conditions. The CA thus mandates the state to effectuate the institutional, material, social and educational conditions needed to place all citizens above threshold levels of capability. Achieving such conditions, however, requires a radical reconfiguration of the ways people work, learn, and live. I propose that a CA-inspired policy agenda can offer a better frame through which to assess AI’s role in this very reconfiguration.

Meaningful work and leisure

Work can – in the right circumstances – enable individuals to access subsistence and produce outputs, while offering a sense of recognition that they’re involved in something valuable. Conversely, the loss of work can also harm individuals in ways that are not strictly monetary but extend to one’s psychological, familial and social status. Mitigating such harm accordingly requires interventions that involve income transfers but go beyond these to establish a comprehensive social security safety net.

Nevertheless, the CA does not value the capability to merely work but to do so ‘as a human, exercising practical reason and entering into meaningful relationships of mutual recognition with other workers’. Beyond liberal interventions, confined to resource distribution and protection from blatant abuse, work should be shaped to support human capabilities. In that regard, AI-driven automation can be desirable to the extent that it is deployed to replace and improve harsh, dangerous or dehumanising types of work. It is undesirable if deployed to manage human work in ways that render the latter more gruelling. Apart from quantitative concerns about the number of jobs lost or created because of AI, policymakers should thus evaluate the qualitative aspects of a partially or fully-automated workplace. 

The CA additionally emphasises the capability for leisure. Leisure provides the temporal and mental space citizens need to engage in political processes and act for their communal well-being, thereby enhancing their agency. So, the state should provide funds and infrastructures for leisure alongside work opportunities that leave room for recreation. Insofar as AI-driven automation might help source the prosperity and free time needed for leisure, it appears desirable. It is instead undesirable if introduced in the workplace for exclusively productivist ends, meaning not for the sake of reducing workers’ assignments but to make them produce more outputs, for instance, by increasing expectations of worker performance and productivity.

More broadly, if AI-enabled productivity gains ensured material well-being, citizens could refuse or reduce work that is not capability-enhancing. AI-funded schemes such as a four-day working week, a Universal Basic Income or a Participation Income could offer them a foundation from which to pursue their valuable doings and beings, inside or outside work. In short, AI may expand (or frustrate) capabilities for meaningful work and leisure. To be effective, this expansion could evolve into a corresponding socioeconomic ‘right’ to meaningful work and leisure.

Practically, policy and technology assessments should examine the degree to which AI-driven automation and related policies:

  • convert productivity growth into expansion of citizens’ capabilities across the board;
  • provide opportunities for meaningful work and leisure and the resources needed to seize these opportunities; and
     
  • replace work tasks/types that hinder human capabilities or are dehumanising.
     

Means vs. ends

Without negating instrumental goods, the CA shifts the focus to the end goal. Accordingly, AI isn’t desirable as such but for its ability to generate shared prosperity and thereby opportunities for meaningful work and leisure. Such prosperity is likewise valuable inasmuch as it advances capability.

This means-ends distinction contrasts with human capital approaches, which evaluate individual abilities based on their usefulness for production and economic growth. The CA doesn’t consider humans a mere means but the ultimate end of production. Therefore, AI should serve and augment human capabilities, avoiding the scenario where workers serve and adapt to AI’s capabilities. This demand is pertinent to the instrumentalisation of micro/crowd/gig workers that fuel AI itself, performing notoriously machine-like work.

Human capital approaches also value education for its contribution to economic growth. However, the training that helps individuals in their role as economic actors doesn’t necessarily help them develop capabilities and flourish. In light of AI-driven automation, educational policy should adopt an orientation broader than human capital formation, including but transcending employability and preparation for the ‘jobs of the future’ (i.e., those that are less automatable).

From anxiety to opportunity

AI is a source of anxiety and concern but might equally be a source of hope and opportunity. A compelling case can be made for a positive vision of less and better work – for the many rather than the few – with the help of AI. Even if such a vision might be deemed utopian by those vested in the status quo, it extends the realm of possibilities beyond the constraints of dominant fatalistic perspectives on the future of work.

Although this blogpost doesn’t provide clear-cut answers, it sketches the potential of AI-driven automation as a reconfiguration able to secure basic levels of prosperity and capability. It assigns policymakers to the task of achieving this reconfiguration and offers a roadmap for a CA-inspired policy agenda. This agenda considers but doesn’t rely on the exact state of AI, and so applies to both the near and distant future of work. It envisions a new target for UK policy, that of a multidimensional growth, concurrently supportive of economic progress, technological innovation and human capability.

Anastasia Siapka is a lawyer and researcher affiliated with the KU Leuven Centre for IT & IP Law (CiTiP) in Belgium, where she also completed her PhD as an FWO Fellow.

This blogpost is based on her paper  ‘Reframing AI Governance: Capabilitarian Insights for the Future of Work’ published on Síntese: Revista De Filosofia.