1.1 About the British Academy
The British Academy is the UK’s national body for the humanities and social sciences – the study of peoples, cultures and societies, past, present and future. We have three principal roles: as an independent fellowship of world-leading scholars and researchers; a funding body that supports new research in the humanities and social sciences, national and internationally; and a forum for debate and engagement – a voice that champions the humanities and social sciences.
The humanities and social sciences have a rich and unique contribution to make to the world we live in. The British Academy’s fellowship represents breadth and excellence across these disciplines, and the Academy’s policy work is dedicated to applying that insight to policy issues for public benefit and societal wellbeing. We bring independence, authority and objectivity to complex issues to enlighten the context, meaning and practicalities of challenges in public policy.
We have an ongoing programme of work on data and AI. This programme asks how big data, data-driven technologies and artificial intelligence (AI) are changing the way that people live, and how we can harness this change for good. A key publication in this programme is the impact of artificial intelligence on work, an evidence synthesis by the British Academy and Royal Society (2018).
1.2 About UCL Public Policy and Grand Challenges
UCL Public Policy supports engagement between a diverse range of researchers and policy professionals in order to enhance the use of evidence and expertise in policy and decision making.
UCL’s Grand Challenges of Transformative Technology and Justice & Equality convene and foster cross-disciplinary research, partnerships, and initiatives across UCL and with external partners. GCTT explores the social impacts of new technology and how data can be used for good. GCJE examines the barriers people face to justice and how societal structures perpetuate and sustain inequalities.
1.3 About this submission
This submission is a summary of discussions that took place at a recent workshop hosted by the British Academy and UCL, as part of a longer project investigating social implications of AI on the Future of Work. This workshop took place over two virtual sessions on Monday 22 and Monday 29 June 2020 and included expert contribution from representatives from parliament, government, academia, business, technology, and education. Discussions were held under the Chatham House Rule. As such, this submission does not constitute a formal policy position of the British Academy nor its Fellows, UCL Public Policy, UCL Grand Challenges or any of the individual attendees of the event, and we do not intend it to represent comprehensive coverage of the issues. However, we believe that it will nevertheless be of interest to the Committee.
The submission has three short sections:
We have also attached two further documents that provide additional information:
It can be very easy to overestimate what people and organisations know about AI. Significant outreach is needed, including to reach people who think that AI is not relevant to them.
The skills required for the better integration of AI into the workplace are not just technical skills in computer science. They also include skills like design, communication, critical thinking and creativity.
The pace of change in AI presents a challenge for education, retraining and upskilling because skills can rapidly become obsolete. We need a focus on adaptable curricula, transferable skills and continuous learning.
The varied ability of large corporations and SME’s to upskill employees and support additional training may be amplified by advancing technology.
Quality and equity
We need to bridge the digital skills and infrastructure gap – providing everyone with basic digital literacy and access to core infrastructure like broadband internet.
Whilst overstated, there is a significant risk of creating an underclass of low-paid, low-skilled work. Avoiding this requires that AI does not exacerbate existing inequalities.
Machines built by people display bias. It is vital that we address this in order to ensure that we do not build inequities into systems that have the potential to become very powerful.
Regional and sectorial disparities will likely be further exasperated by advances in technology, its availability and the skills to take advantage of it.
2.2 What don’t we know about how AI might impact future skills and the quality and equity of work?
How might the barriers to an AI-related career increase as AI gets more complex?
What exactly are the skills gaps in the UK economy? Where are they located – by sector and by geography? Which skills might be most in demand in the future?
How do we weigh the importance of highly technical skills against more generalist ‘soft’ skills in future training?
How is retraining or upskilling later in life different for people with more or fewer digital skills and competencies?
How do we make better use of national assets in order to foster investment in life-long leaning?
What can we learn from other countries and AI skills initiatives?
How does the UK digital skills agenda evolve after exiting the EU?
What kinds of smart regulation does the UK need to enable action on upskilling and life-long learning?
Quality and equity
What kind of society do we want to live in?
What constitutes equitable work ?
How can we involve people with very little interest or skill in technology in conversations about technology policy?
How can we best communicate the benefits, risks and uncertainties of AI? How can we best stimulate informed public debate about the trade-offs? Who has responsibility for engaging the public in this debate?
What will the impacts of COVID-19 be?
What will the impact of exiting the EU be?
2.3 What should we do to advance action on AI and the future of work?
Employers, educators and government (at all levels) all have a role to play in advancing action on AI and the future of work that in a way that puts future skills, quality and equity at the heart. Some of these roles are explored below.
Educators need to be able to train learners to work with AI, and to understand its applications. This includes a wide range of jobs that do not yet exist. This requires education and training that focuses on adaptability, critical thinking skills, and how to learn.
The increasing ubiquity of AI will mean many more ‘low code’ and ‘no code’ jobs that involve AI. This requires education and training to ensure basic digital literacy across the population, and a better appreciation of the importance of non-coding skills like communication, problem solving and design.
Policymakers need to carefully consider the possible roles for primary, secondary, tertiary and continuing education.
Most businesses want to innovate and upskill their workforces, but there is huge variety in their ability to do so. Strategies for supporting SMEs, including those which think that AI is irrelevant to them, will need particular thought. Strategies could include working closely with membership organisations, or looking at clusters of industry in particular regions.
Workers will need to retrain and upskill throughout their working lives. This should be well-integrated into work, and educators and employers should work together. Workers should be empowered not only to develop their skills, but also to adapt their workplaces inline with the skills that they have gained.
Businesses should see improving the AI skills of their workforce and their uses of AI more widely as driven by their business purpose, rather than as part of a corporate social responsibility strategy.
There may be a significant role for government (at all levels) in coordinating and bringing together educators, employers and individuals across sectors and places to develop shared solutions.
There may also be a role for government in ensuring a basic standard of access that will underpin a positive future for AI and work. This could include minimum digital literacy standards and access to infrastructure including broadband internet.
AI is only one of the factors that will cause major changes in the nature of work over the coming decades, and it is not the first time that work has undergone a transformation related to technology. For example, changes due to AI will sit alongside changes related to climate change, and many systems have already undergone significant shifts from paper to digital systems. The role of AI should be considered in this wider context of overlapping, interdependent factors in order to ensure an equitable transformation.
There is a role for public culture and the arts in provoking and enabling a conversation about AI, encouraging people from all backgrounds and workers from all industries to contribute their thoughts.
We would be very pleased to speak with you further about any element of our response. We expect to be using the outputs of this workshop discussion to develop a larger programme of work and would be pleased to keep the Committee updated as this develops over the summer.
Data and AI, The British Academy
UCL Public Policy
UCL Grand Challenges
AI and the Future of Work
A Cross-Disciplinary Workshop
10.00am-12.00pm, 22 and 29 June 2020
About the workshop
The widely acknowledged digital skills gap holds potential for profound impact on future economic growth. Enabling and upskilling the workforce to take full advantage of the digital and AI technology revolution will be vital in both a post-Brexit and post-COVID-19 world. However, there remains little consensus on the ways that AI should intersect with work, or the place of Ai in the wider political, economic and social discourse. Likewise, questions remain as to how Government will be able to support the investment in lifelong skills and training that will be required to shape AI for the benefit of all.
To begin to unravel these challenges, UCL and the British Academy held a two-part, virtual workshop to identify how researchers, policy professionals, employers and training providers can respond to the changing nature of work and support the labour market.
The workshop built on the impact of artificial intelligence on work, an evidence synthesis by the British Academy and Royal Society (2018), and addressed several areas of research interest (ARIs) identified by different government departments.
About these notes
These notes summarise discussions at two roundtable events hosted by the British Academy and UCL Public Policy and the UCL Grand Challenge of Transformative Technology on Monday 22 and Monday 29 June 2020. This document is not intended to represent the views of the British Academy or UCL, nor does it represent the views of individual attendees of the event.
Part 1: What do we know, and what do we not know?
Purpose of Part 1 and background briefing paper
Dr Jack Stilgoe, UCL
The aim of this workshop is to get a picture of what we know and what we don’t know about the ways in which AI and work interact. We are not presuming that the causal relationship between advances in technology and changes to work is simple or unidirectional. Rather, we are interested in understanding how they are entangled, and in identifying gaps in that understanding.
In order to support discussion, all participants received a copy of a background briefing paper drafted by Em O’Sullivan, a PhD student in the Department of Science and Technology Studies, UCL. The briefing paper takes a wide range of information about AI and the future of work, and identifies a set of questions to enable a sharp discussion. Today’s discussion will focus on two of the sections identified in the paper:
Our aim is to address these questions, and in the process to tease them apart, expand them, and add new questions. Some of the directions that we might take the discussion in include:
The briefing paper also includes a discussion on roles and responsibilities. This will be addressed in the second part of this workshop, next week. In that discussion, we will also look at more radical propositions that respond to advancing technology, such as a universal basic income or robot taxes.
The impact of artificial intelligence on work
Jessica Montgomery, The Royal Society
In 2017, the Royal Society convened a series of public dialogues on AI and the future of work. In these dialogues, two visions of AI and the future of work emerged, each occupying an extreme: AI will either be the end of employment, or it will enable a utopian society in which work problems are solved. In 2018, the Royal Society and the British Academy commissioned an evidence review to test the strength of the evidence behind these two extreme predictions.
The resulting publications, the impact of artificial intelligence on work, found that the evidence suggests that neither prediction is likely. Instead, it is much more likely that AI will have a disruptive effect on work – some jobs will be lost, some will be create, and others will change. It found that technology is not a unique or overwhelming force, and we can expect political, economic and cultural factors to all shape what type of change we see. Learning from the history of technological advances, the evidence synthesis also found that while technologies generally contribute to an increase in population-level productivity, employment and economic wealth, these benefits are only felt over quite long timelines. In the transition period there is disruption, and some people lose out.
The evidence synthesis also looked at the kinds of policy interventions that have been proposed in relation to AI and work. These include:
Since the publication of the evidence synthesis, we have seen an evolving public discussion about AI. In 2017, we were at the peak of the hype cycle, with press coverage of AI focused on extreme scenarios. Now, in 2020, there has been a significant shift towards more nuanced discussions of how AI will change workplaces, through, for example, bias or surveillance. We have also seen an increased interest in international comparisons, and the UK Government’s AI Sector Deal, which has ‘good jobs and greater earning power for all’ as one of its five foundations of productivity.
Since 2018, we have also seen significant changes in our wider political and social context, including the Government’s ‘levelling up’ agenda and the current COVID-19 pandemic, both of which have the potential to have a significant impact on technology policy.
What do we know?
What do we know about how individuals view the challenges and opportunities of advancing AI technology for decision-making about careers and skills development?
What don’t we know?
What don’t we know about how individuals view the challenges and opportunities of advance AI technology for decision-making about careers and skills development?
Quality and equity
What do we know?
What do we know about how AI might impact the quality, equity and suitability of work?
What don’t we know?
What don’t we know about how AI might impact the quality, equity and suitability of work?
Part 2: What should we do?
Summary of Part 1 and purpose of Part 2
Professor Rose Luckin, UCL
The first part of the workshop focused on the questions of what we know and what we don’t know about AI and the future of work (summarised above). We identified a wide range of useful questions, focusing on future skills and quality and equity. Most importantly for the second part of the workshop were questions such as:
The second part of this workshop will consider these questions, asking what should be done to address them, and whose responsibility it is to address them, considering geographical and sectoral variations.
The wider political and policy landscape
Anna Bradshaw, The British Academy
Our conversations today take place in the wider political and policy context. Four key issues in that wider landscape to keep in mind during our discussions are: (1) the policy ambitions of the (still relatively new) government, including ‘levelling up’ and, possibly, changes to the civil service; (2) the ongoing process of leaving the EU; (3) the current COVID-19 pandemic and the coming recovery and expected recession; (4) major global protest movements including the Black Lives Matter movement and the youth strike for climate.
The role of employers
Rob McCargow, PWC
A recent survey by PWC of over 20,000 adults in 11 countries found more than half expect AI to change their job, and over 60 per cent are positive about the impact of technology on their work. However, while most (77 per cent) respondents would re-train in order to improve their employability, only one third are given the opportunity to develop their general digital skills. It is the responsibility of employers to upskill their workforce, providing opportunities to their employees; PWC has looking to upskill its entire workforce. However, the unilateral action of individual businesses will not be sufficient. We can look to examples of good practice, like the Digital Skills Bridge in Luxembourg that brings together businesses, government and third sector organisations to develop a national strategy and a support mechanism for employers.
The role of education
Vanessa Wilson, University Alliance
One of the key purposes of education is to prepare learners for work. This purpose will be of increasing importance in the short and medium term as we move out of the immediate COVID-19 crisis into a possibly terrible recession, and in the long term as we see an ever-increasing role for technology at work. Key questions to ask include: (1) At what level – primary, secondary, tertiary – should preparation begin? (2) How do the curricula and teachers’ skills keep pace with developing technology? (3) How can we equip current workers to survive and thrive? (4) Whose responsibility is it to ensure that the workforce of the future is appropriately skilled? (5) How can we make AI an technology attractive to current and prospective students? (6) How can we ensure that technology is harnessed to level up and create parity of opportunity, instead of repeating or even exacerbating the inequities of the past?
Policy and education
What might be the responsibility of different groups in advancing action on AI and the future of work related to education?
Policy and employers
What might be the responsibility of different groups in advancing action on AI and the future of work related to employers?
Policy, education, and employers
Policy and individuals
What might be the responsibility of different groups in advancing action on AI and the future of work outside of educators and employers?
Reflections and next steps
Dr Jack Stilgoe, UCL
This discussion has, in many ways, made the questions we were asking helpfully less clear. The roles and responsibilities of policy, educators, employers and individuals are complex and overlapping. It reminds me of a quotation from Carl Sagan that I use with my first-year students:
We’ve arranged a global civilisation in which most crucial elements profoundly depend on science and technology. We have also arranged things so that almost no-one understands science and technology.
While many people take this quote to mean that we need to become better scientists, I don’t think that Carl Sagan would have agreed. Rather, better understanding science and technology also means understanding a whole range of issues, from data quality and the hidden labour of AI to bias, equity, and ethics. What today’s discussion has clarified is that our approach to understanding AI and the future of work needs to be interdisciplinary and inclusive.
Our current moment of crisis presents a risk and an opportunity. In order to maximise the opportunity, we need to do two things. Firstly, we need an approach that is interdisciplinary, intersectoral, adaptable and thoughtful. Secondly, we need to know where we are going. We need to know what good work in a world with ubiquitous AI looks like.
UCL and the British Academy will take all of the discussion over this workshop to start to flesh out a larger project that, we hope, will begin to do these things, and to answer some of the fascinating questions we have raised
There is a widely acknowledged digital skills gap in the UK, with the potential for profound impact on the future growth of the UK economy. Enabling and up-skilling the UK workforce to take full advantage of the digital and artificial intelligence (AI) technology revolution will be vital in both a post-Brexit and post-COVID world, however there remains little consensus on the impact of AI on the nature of work in the UK, as well as its place in the wider political, economic and social discourse.
Likewise questions remain as to how Government will be able to support the investment in lifelong skills and training that will be required to harness this opportunity to its full for the entire UK.
The UK’s Industrial Strategy names AI & Data as one of its four Grand Challenges(1). The challenge of AI includes the questions:
There are many possible “futures”(2) of AI and work, and policy will play a key role in shaping the future work landscape of the UK. There is a need for researchers, policy makers and industry to pro-actively guide AI policy to meet this challenge.
This briefing paper provides a review of current evidence from academic and policy literature around AI and its potential impact on the future of work in the UK.
How do individuals view the challenges and opportunities of advancing AI technology for decision making about careers and skills development?
While many people are aware of some everyday technologies that use AI, such as personal assistants on smart phones and targeted online advertisements, very few people are familiar with key AI terms such as “machine learning” or have a good understanding of how AI technologies work(3). Sensationalist depictions of AI in the media, along with a lack of clarity from AI developers about what their technologies can and cannot do, has fed some of this confusion(4).
Decision making about careers and skills development is also strongly influenced by social factors such as gender, ethnicity, socioeconomic background and age.
The reasons behind technology skilling decisions are complex. Initiatives to reduce the AI digital skills gap will need to be informed by wider social research into technology engagement and the issues faced by members of under-represented social groups.
How might advancing AI impact the quality, equity and suitability of work?
Advances in automation capabilities as a result of AI risk the loss of jobs in certain employment sectors and the potential eradication of some occupations altogether. A general consensus has emerged on what jobs are most at risk in the next 10-20 years:
This raises considerable equity issues as these projections indicate that the benefits of AI and automation will not be spread equally across different segments of society, and its negative impacts are likely to be felt disproportionately by people who are already in relatively low-income occupations. Some possible policy interventions to reduce inequity are raised later.
What is less well understood is how the quality of future work may be affected. Since developments in AI will create new jobs that don’t yet exist and radically alter existing jobs in unforeseen ways, to some extent it is impossible to predict the landscape of the future labour market. Discussions of some of the possible impacts range from optimistic to critical:
What are the roles and responsibilities of Government, employers and educators in improving outcomes for individuals and society to meet the evolving work landscape?
The question of roles and responsibilities is the least understood area in the current literature. While there is a consensus that there is significant risk of increased economic inequality if the benefits of AI are not redistributed throughout society, many diverse and occasionally conflicting strategies have been suggested for achieving this. This section briefly introduces some of these potential avenues to prompt further discussion.
The responsibility to boost productivity and economic growth:
The responsibility to reduce inequality and ensure quality of life:
The responsibility to provide legal protection for groups and individuals:
The capabilities of AI continue to advance rapidly. While this briefing paper provides a snapshot of research at this point in time, an ongoing assessment of the landscapes of the labour market and the AI industry will be required in order to inform policy.
Em O’Sullivan, PhD Candidate, Science and Technology Studies, UCL
1. Department for Business, Energy and Industrial Strategy, Industrial Strategy: Building a Britain fit for the future (November 2017).
2. S. Ojanperä, N. O’Clery, & M. Graham, Data science, artificial intelligence and the futures of work (October 2018).
3. Ipsos MORI, Public views of machine learning: Findings from public research and engagement conducted on behalf of the Royal Society (April 2017).
4. Select Committee on Artificial Intelligence, AI in the UK: Ready, willing and able? (April 2018).
5. HESA, ‘Table 9 - HE student enrolments by subject of study 2014/15 to 2018/19’, https://www.hesa.ac.uk/data-and-analysis/students/table-9.
6. The British Academy & The Royal Society, The impact of artificial intelligence on work: An evidence synthesis on implications for individuals, communities, and societies (September 2018).
7. C. B. Frey & M. A. Osborne, ‘The future of employment: How susceptible are jobs to computerisation?’ in Technological Forecasting & Social Change, vol. 114 (2017), pp 254–80.
8. Matthew Fenech, Cath Elliston, and Olly Buston, The impact of AI in UK constituencies: Where will automation hit hardest? (October 2017).
9. W. Edward, ‘The Uberisation of work: The challenge of regulating platform capitalism. A commentary’ in International Review of Applied Economics (2020).