Yes, AI could boost productivity, but work is about more than maximising output

Worries about the British economy have long been dominated by one persistent concern – weak productivity. Since the financial crisis of 2008, growth has stagnated, leaving the UK trailing well behind the US, France and Germany across that whole period.

One familiar response to this problem is to suggest that if the British workforce could somehow produce more in less time, prosperity would follow and all would be well. New technology, particularly AI, is often presented as the solution.

The UK government certainly seems to like the idea, placing AI and technological innovation at the centre of plans to boost economic performance. At a speech to business leaders on March 17, chancellor of the exchequer Rachel Reeves promised £2.5 billion of investment in AI and quantum computing to get things moving.

But what if productivity is not the problem we should be solving?

Increasing the country’s “output per hour” – the unit by which productivity is measured – does not necessarily make work more secure, more fairly rewarded or more socially useful. And nor does it make the UK more economically resilient.

In fact, it can do the opposite. Prioritising efficiency to boost productivity – by cutting costs and relying on tightly configured supply chains – can make economic systems extremely fragile.

Productivity problem

The problem with focusing too much on productivity is most obvious in some of the sectors that are central to our day-to-day lives. The effectiveness of care work, healthcare and education, for example, all depend on human interaction.

But teaching a class, caring for an elderly person or treating a patient require time, attention and professional judgment, making it difficult to increase “output” in the same way as in more automated sectors. There are limits to how much faster a nurse or teacher can work without undermining the quality of what they do.

Economists have long recognised that services which depend on human interaction – referred to as being “labour intensive” – face limits to productivity growth, because many of the tasks involved cannot be significantly sped up or automated without affecting quality.

This dynamic is referred to as “Baumol’s cost disease” – an economic theory which shows that costs will inevitably rise over time in labour-intensive sectors, despite little or no productivity growth.

Yet these sectors are essential to long-term social wellbeing and economic stability. They sustain everyone’s health, skills and security.

Another issue with increasing productivity comes down to the fact that for quite some time, the UK economy has been heavily weighted towards areas like finance, education and the creative industries. Manufacturing plays a much smaller role.

But in manufacturing, technological improvements can translate more directly into higher output per worker. This is what happens when industrial robots automate assembly-line tasks, allowing a single worker to oversee machines producing far more units than manual labour alone could achieve.

In contrast, much of the work undertaken in the UK, from management to care, depends on interaction, judgment and time. Its value is real but not easily measured.

The UK is therefore trying to solve a productivity problem in sectors where productivity is inherently difficult to define and improve.

Alternatives to output

This in turn points to a broader issue. The future of work is not just about how much we produce, but about how work is organised, how its rewards are shared, and how it fits into the rest of life.

None of this means productivity should be ignored – but it is a narrow measure. When treated as the primary goal of economic policy, it can produce an economy that appears efficient on paper yet fragile in practice, with rising output alongside stagnant living standards.

This was evident in the UK after the global financial crisis, when employment and GDP recovered while real wages stagnated for much of the 2010s. Productivity growth alone does not guarantee broadly shared prosperity.

The UK’s productivity slowdown is often framed as a failure to generate enough output per worker. A more uncomfortable possibility is that it reflects a mismatch between what the economy measures and what society needs.

Technology like AI may increase what workers can produce in an hour. But if the problem lies in how work is organised and valued, greater efficiency alone will not be enough.

Questions about the future of work should not begin with productivity statistics alone. They should begin with a simpler inquiry: what do we want the work we do to achieve in the first place?

Abigail Marks received funding from UKRI Rapid response COVID funding