Gauging the impact of artificial intelligence on work

This article is published in British Academy Review No. 34 (Autumn 2018).

The print version of this article can be downloaded as a PDF file.

Genevra Richardson is Professor of Law at King’s College London. She was elected a Fellow of the British Academy in 2007, and is the Academy’s Vice-President (Public Policy).

Through our public policy work, the British Academy aims to use both the expertise that exists within our Fellowship and the findings from the research we fund to provide policy-makers with insights into some of society’s greatest challenges.

It takes many minds to respond to the complex and interconnected questions that face our society, and we must approach these important questions with intellectual rigour and draw on a wide range of academic disciplines. Many of these great questions – such as the impact of technological change – are as much human and societal challenges as they are scientific.

It is difficult to think of a more important set of questions than how society can successfully realise the benefits, and manage the disruption, from the technological transformation that is currently taking place. The need to respond to technological change isn’t new, but what might be different this time is the sheer pace at which technologies, such as artificial intelligence, are developing.

This represents a real challenge for policy-makers, as does the wide range of predictions and projections that tend to proliferate in this area. For these reasons, the British Academy decided to work with the Royal Society to publish a synthesis of the evidence that exists for the impact of artificial intelligence on work.

Our work reviewed and synthesised evidence from across many different academic disciplines, in order to inform policy debates about the interventions needed, and to support a more nuanced discussion about the impact of AI on work. What we were trying to do was to get behind the hype that Margaret Boden quite rightly identifies (p. 00), and to discover how much certainty we can derive from current evidence.

While many of the public and policy debates on AI and work tend to swing from apocalyptic fears of the ‘end of work’ to reassurances that there will be little change in overall employment rates, the evidence suggests neither of these extremes is likely.

History demonstrates that in the long run new technologies do increase population-level productivity, employment and economic wealth. But these benefits take time to emerge, and there can be significant periods in the interim where some in society are very negatively impacted, as the article by Jane Humphries and Benjamin Schneider so vividly illustrates (p. 00). Evidence from both historical and contemporary sources suggests that technology-enabled changes to work tend to affect lower-paid and lower-qualified workers more than others.

Where there is less certainty is around the exact number of jobs likely to be lost, gained or changed by AI, although there does seem to be an emerging consensus that around 10-30 per cent jobs in UK are highly automatable.

The future is not set in stone, and AI and our response to it could go down a number of possible paths. There are choices to be made, and these could be aided by the sort of ‘outcome-thinking’ that Richard Susskind is advocating.

But to make these choices will mean continuing to use the very best available evidence from across a wide range of academic disciplines, including the humanities and social sciences, so that as a society we can be confident that technological change is harnessed in a way which helps improve productivity, ensures this benefit is shared across society, and enables humans to flourish.