By Diane Coyle
If you spend less than four hours and 20 minutes a day online, you are below average for the UK. Using apps and online services, many of them free, is part of the fabric of daily life for most of us. Most organisations similarly have embedded digital technology in the way they operate. For many businesses it has changed their business model. And while the digital transformation of the economy – and the rest of life – has been under way for a good 30 years, AI is ratcheting up the change.
The consequences of the digital and AI revolution will be as wide-ranging as those of previous ‘general purpose technologies’ such as printing or electricity. A full evaluation of the consequences for society, from the joy of chatting to a family member across the world to the tragedy of many online harms, is impossible. My focus is on the narrower economic questions: how can AI be shaped to achieve good outcomes? And how should we be monitoring its economic effects?
The definition of a general purpose technology is that it spreads through the whole of the economy with broad effects, so this is a challenge. What’s more, the AI tools most people in the UK will use will either be provided by large US companies or depend on their services, so – like policymakers in most other nations – the government’s influence will be limited. In the current context there are even political limits on how the UK can regulate AI. But what does lie in our own hands is how the technology is used.
The obvious starting point is to consider the possible impact on jobs. There has already been research on this question, but it is inconclusive: AI might replace human workers, or it might augment them. Some businesses will undoubtedly go down the former route, trying to cut costs. In other cases, employees will be able to use AI to enhance their work and become more productive. In this respect, AI is similar to the many previous generations of automation technology, so the results for people’s employment and livelihoods will be uncertain.
This points to a clear need for policies that help people navigate the change – something UK governments painfully failed to do in the early 1980s and 1990s as manufacturing industry automated. At a minimum people will need help a less rigid education focusing on distinctively human strengths, reskilling, access to a more flexible benefits system, good information and guidance – obvious points but not areas of policy where governments have previously proven themselves capable.
We should also acknowledge the emotional impact of changing work. In an important report on AI forming part of the Pissarides Review, the Institute for the Future of Work has emphasised the potential stress and adverse impact on people’s wellbeing – both from the job market uncertainty but also from the growing use of AI to organise and monitor people’s work. These developments may call out for changes to employment legislation in future.
Beyond labour market changes, the economy also stands at a crossroads in terms of people’s experience as citizens and consumers. There is great potential for AI to improved productivity and service. The Government clearly sets great store by the hope that it can deploy the technology to make public services more efficient, as do many businesses. AI could speed up many processes, and enable workflows to change to enhance productivity and quality, as did earlier generations of digital technology.
However, gains like these will not be automatic. It is expensive and disruptive to re-engineer business processes. Finance directors (and the Chancellor) will be more interested in cost savings than quality improvements. Rather than speedier and more personalised services, we might find ourselves facing chatbots unable to respond to non-standard queries and the kind of tax on our time that many online suppliers have already introduced through frictions and ‘dark patterns’ online. The automatic checkouts all supermarkets feature offer a caution: they have saved labour and made stores more productive (as it is measured) but the shopper’s free work substitutes for the previously waged employee.
Which leads to a final point about how to measure and assess the changes under way. The statistical system is already unable to monitor effectively the digital economy, including business model changes, making it hard to assess the extent of productivity gains or any broader measures of progress. Without suitable measures even of changes in the labour market or organisations’ adoption of AI, policymakers will be unable to shape this powerful technology in ways that benefit society as a whole.
The digital economy and AI highlight the limitations of our over-reliance on GDP as a single measure of progress. AI will almost certainly boost economic output over the coming years, but it should be the job of government to ensure it does so in a way that we can all benefit from. That obligation could be clearer if we were guided not by a single gauge but instead by a whole new framework of metrics that takes into consideration current economic realities.
Diane Coyle is the Bennett Professor of Public Policy at University of Cambridge. She is on Bluesky: @dianecoyle1859.bsky.social
This article is the fifth in PBE’s new ‘Economics to improve lives’ series that explores a question at the heart of PBE’s mission: How do we ensure that wellbeing – the quality of life experienced by individuals – is the ultimate goal of government?
We’re bringing together thinkers and commentators from across economics, policy, academia, media and civil society to challenge conventional wisdom and consider how we might build an approach to measuring economic success that puts the lived realities of people at its heart and fits the times we live in. Opinions are the author’s own.
Read previous articles in this series from Hetan Shah, Nancy Hey, Sarah Davidson, Jon Franklin, Ed Humpherson.