192.The economic impact of AI in the UK could be profound. This chapter considers two widely shared concerns for policymakers, businesses and the general public: the UK’s productivity puzzle, and the potential impact of AI on the labour market.
193.The opportunity that the widespread use of artificial intelligence offers to improve productivity in the UK was, perhaps, the most common benefit cited to us by our witnesses. Andrew de Rozairo, Vice President of Customer Innovation and Enterprise Platform, SAP, said “if we adopt AI, given the strong skillsets that we have in the UK, we have a huge opportunity to boost productivity”. TechUK told us: “it is likely that the adoption of AI by companies will increase productivity, efficiencies, cost savings and overall economic growth across all industries and sectors”. The Government also recognised this benefit: “Impacts in industry … are likely to be profound in terms of productivity”. A number of witnesses argued that productivity would be improved as human labour was augmented in a range of ways, such as summarising complex documents or sorting email inboxes.
194.However, a note of caution is advisable, as the economic consequences of earlier phases of computerisation, particularly in the 1970s and 1980s, are still poorly understood. Some economists have argued that there is still little evidence that information technology had a significant impact on productivity over this period, especially in the United States, a phenomenon which famously led economist Robert Solow to remark that “you can see the computer age everywhere but in the productivity statistics”. Others argue that increased productivity in the 1990s show that gains were merely delayed. Either way, prior experiences with computerisation suggest that any relationship between AI adoption and productivity is unlikely to be necessary or straightforward in nature.
Productivity measures how efficiently work is converted into the output of goods and services. The better the productivity, the more goods and services are being produced per hour worked. Productivity is used to assess economic growth and competitiveness, and as the basis for international comparison of a country’s performance. In the UK, the Office for National Statistics regularly reports on labour productivity, based on output per hour, output per job, and output per worker.
195.Hopes that AI will improve productivity must be set against a backdrop of low productivity growth across the developed world, and almost non-existent productivity growth in the UK since the 2008 financial crisis. The Royal Society for the encouragement of Arts, Manufacturers and Commerce (RSA) told us of “lacklustre productivity levels, with UK workers on average 35% less productive than their counterparts in Germany and 30% less productive than workers in the US”. Sarah O’Connor said this was “a puzzle that is taxing the best minds in technology and economics right now”. She also told us that “if you do not have productivity growing at a decent clip then you cannot have sustainable increases in living standards” and AI “could mean a step change in productivity”. The Center for Data Innovation put it in even more stark terms, saying that finding a way to improve productivity was:
“ … particularly critical for the UK, which is suffering from an unprecedented productivity crisis, with productivity stagnant over the last decade. Unless Britain can find a way to boost productivity, social and political crises will increase as incomes stagnate”.
196.The Office for National Statistics reported in January 2018 that productivity had grown by 0.9% from Quarter 2 (April to June) 2017 to Quarter 3 (July to September) 2017—this is the largest increase in productivity since Quarter 2 2011. This is, however, not a significant enough improvement to overlook the potential benefits that AI might have for productivity in the UK.
197.We share the optimism of our witnesses that AI could improve productivity. We also share their concerns that this is an opportunity that could be missed in the UK. The Institute of Chartered Accountants in England and Wales (ICAEW) said: “we must not lose sight of the reality of most businesses, who are a long way behind in their adoption of many technology trends, including AI”. Sage told us that their research demonstrated “that companies currently spend an average of 120 working-days per year on administrative tasks. This accounts for around 5% of the total manpower for the average small and medium-sized business”. The research suggested that the amount of time spent on such tasks is because digital tools have not been adopted, and that if UK businesses could be 5% more productive, GDP could increase by £33.9 billion per year. Kriti Sharma told us that “lack of digital adoption is leading to what we call the productivity gap in the UK”.
198.In 2017, 5.7 million businesses in the UK were classified as SMEs (99% of all businesses), with 5.4 million of those employing fewer than 10 staff. Such micro-businesses accounted for 33% of employment and 22% of turnover. The CBI, which represents over 190,000 businesses in the UK, said: “Digital innovations are at the heart of economic, social and cultural development across the UK. They drive productivity, help to raise living standards and lay the foundations for tomorrow’s world”. If businesses are not using existing technology, in particular SMEs, we are concerned that the potential benefits to productivity offered by artificial intelligence could bypass significant portions of the business community in the UK.
199.We support the Government’s belief that artificial intelligence offers an opportunity to improve productivity. However, to meet this potential for the UK as a whole, the AI Council must take a role in enabling AI to benefit all companies (big and small) and ensuring they are able to take advantage of existing technology, in order for them to take advantage of future technology. It will be important that the Council identifies accelerators and obstacles to the use of AI to improve productivity, and advises the Government on the appropriate course of action to take.
200.Other witnesses shared concerns about the state of digital infrastructure in the UK. Research Councils UK said:
“Localities with lower levels of investment in technological and digital infrastructure and low skill levels are likely to be hardest hit by AI technologies. Investment is needed to access the rewards of adoption of AI”.
201.Vishal Wilde said “those who live in rural areas and who do not have access to broadband also do not feel the benefits of … the productivity gains associated with AI nearly as much as other parts of the country”. Artificial intelligence is, in part, reliant on access to digital infrastructure. Without the digital foundations (both physical in terms of internet connectivity and in terms of skills, discussed later in this report) the potential benefits of artificial intelligence to the UK’s productivity will be neither realised nor widespread.
Superfast broadband is defined, by Ofcom, as connections providing download speeds in excess of 30 megabits per second (Mbps).
Ultrafast broadband is considered to be where speeds are in excess of 300 Mbps.
202.To improve national infrastructure, the Industrial Strategy sets out plans to increase the National Productivity Investment Fund from £23 billion to £31 billion, and to improve digital infrastructure with over £1 billion of public investment, including £176 million for 5G and £200 million for full-fibre broadband networks. Matt Hancock MP told us that 95% of premises would have access to superfast broadband by the end of 2017, and that ultrafast connectivity rollout is the next ambition for the Government. This would be delivered via “a competitive market with many players bringing ultrafast speeds over full-fibre technology”. As of February 2018, 3% of the UK was covered by full-fibre broadband.
203.We welcome the Government’s intentions to upgrade the nation’s digital infrastructure, as far as they go. However, we are concerned that it does not have enough impetus behind it to ensure that the digital foundations of the country are in place in time to take advantage of the potential artificial intelligence offers. We urge the Government to consider further substantial public investment to ensure that everywhere in the UK is included within the rollout of 5G and ultrafast broadband, as this should be seen as a necessity.
204.Our witnesses reminded us of the importance of government as a customer, nationally and locally. It can both procure AI solutions for the public sector and adopt the technology, thereby supporting UK-based technology companies.
205.BSA (The Software Alliance), a global software advocate, said: “The UK Government could help demonstrate AI’s potential benefits by investing in innovative AI implementations in the public sector”. Professor Susskind told us that “in the public sector … use of AI and other advanced technologies should transform and not simply streamline our current ways of working and governing”. Doteveryone said: “there is huge potential for improved efficiency in Government and the public sector if AI is used effectively, which would lead to huge savings in public money”. TechUK suggested that “the use of AI virtual agents across Government departments and the public sector could save an estimated £4 billion a year”. Microsoft argued that deployment of artificial intelligence in the public sector could enable more informed policy decisions, and innovative uses of AI could help address public and societal challenges.
206.The Hall-Pesenti Review recommended that:
“Government, drawing on the expertise of the Government Digital Service, the Data Science Partnership and experts working with data in other Departments, should develop a programme of actions to prepare the public sector and spread best practice for applying AI to improve operations and services for citizens”.
207.The Government shared with us as part of their written evidence the AI tools and programmes it is using, or looking to use, in the near future. We welcome the fact that Government departments are actively considering the use of artificial intelligence in the delivery of public services, in particular using innovative approaches such as Kaggle competitions. Other governments are already doing this. For example, in December 2017, the US Department of Homeland Security offered $1.5 million in prizes for their ‘Passenger Screening Algorithm Challenge’, which aimed to improve the accuracy of threat prediction algorithms used in airport security. Datasets were made available and the challenge ran for about a fortnight. Such innovative crowdsourcing can help governments and businesses tap into world-class expertise which they would otherwise not be able to access.
208.However, the Government could still do more to deploy AI and we are conscious that it is considering this. The Autumn Budget in 2017 announced the establishment of the GovTech Catalyst. This will be a small unit based within the Government Digital Service which will provide a direct access point to Government for businesses and innovators. The GovTech fund is £20 million over three years to support public bodies in procuring innovative products. Matt Hancock MP told us “most of GovTech is about procurement … Indeed, getting procurement rules right is one of the most important parts of driving improvements in technology through government, because you need the leadership and the permission from the top to drive the change”.
209.The Royal Society agreed with the Minister’s assessment:
“One direct way in which governments can potentially help start-up companies, where appropriate and allowable, is through their procurement processes. Government contracts help early-stage companies in several ways: they provide a source of income; they give the company the direct experience of engaging with customers, which provides important feedback for their developing market offering; and they act as external recognition of the company’s product”.
210.The Government spends £45 billion a year on procuring goods and services. As such, it is one of the most significant ‘customers’ in the United Kingdom, and has immense power in encouraging the adoption of new behaviours and practices in its supply chains.
211.In the UK, the Crown Commercial Service, an agency of the Cabinet Office, is responsible for procurement policy for the Government, enabling cost-efficient procurement by bringing together requests for the same goods or services, as well as supporting smaller projects. The Crown Commercial Service works with over 17,000 customer organisations in the public sector and has more than 5,000 suppliers, and thereby has significant influence over various sectors in the country.
212.Our witnesses suggested that Government procurement could be used to encourage greater adoption of artificial intelligence, both through the companies contracting directly with Departments and via the Crown Commercial Service. SCAMPI, a research project at City, University of London, said: “the UK public sector … is currently benefiting little from the development and use of artificial intelligence, as few initiatives have been funded or reported”. The UK Computing Research Committee said “the public sector could do more to benefit from these techniques to support the provision and optimisation of services across a host of areas”.
213.Public procurement in the UK is subject to the Treaty on the Function of the European Union’s (TFEU) principles of non-discrimination, the free movement of goods, the freedom to provide services and the freedom of establishment. This is realised via a series of directives, which have been translated into domestic law. In the UK, this means that central Government, and other public organisations, must advertise contracts for goods, services and works which are worth over £10,000 at a UK-wide level, and at an EU-wide level for services and supplies contracts worth over £118,000. The threshold for advertising ‘works’ contracts across the EU is £4.5 million. Given the UK’s departure from the European Union, there is an opportunity for the Crown Commercial Service to ensure that these rules and thresholds benefit businesses in the UK, in particular when it comes to public sector procurement and the stimulation of a fertile AI development sector, as long as it is still a competitive process.
214.In 2013, the Government Digital Service launched the Service Design Manual, intended to help improve public services and ensure the adoption of digital approaches wherever possible. The Manual includes guidance and instructions on how to approach choosing technology, and use the Technology Code of Practice as part of the spend control process. The Technology Code of Practice includes points such as “use cloud first” and “make better use of data”. There is no explicit mention of artificial intelligence in the Code. For the Government to adopt artificial intelligence in the delivery of public services, the need to consider AI solutions must be embedded in the decision-making process right from the start. As such, amending the Code could be one approach to ensure that technologists in the civil service actively consider the use of artificial intelligence.
215.The Government’s leadership in the development and deployment of artificial intelligence must be accompanied by action. We welcome the announcement of the GovTech Catalyst and hope that it can open the doors of Whitehall to the burgeoning AI development sector in the UK. We also endorse the recommendation of the Hall-Pesenti Review aimed at encouraging greater use of AI in the public sector.
216.To ensure greater uptake of AI in the public sector, and to lever the Government’s position as a customer in the UK, we recommend that public procurement regulations are reviewed and amended to ensure that UK-based companies offering AI solutions are invited to tender and given the greatest opportunity to participate. The Crown Commercial Service, in conjunction with the Government Digital Office, should review the Government Service Design Manual and the Technology Code of Practice to ensure that the procurement of AI-powered systems designed by UK companies is encouraged and incentivised, and done in an ethical manner.
217.We also encourage the Government to be bold in its approach to the procurement of artificial intelligence systems, and to encourage the development of possible solutions to public policy challenges through limited speculative investment and support to businesses which helps them convert ideas to prototypes, in order to determine whether their solutions are viable. The value of AI systems which are deployed to the taxpayer will compensate for any money lost in supporting the development of other tools.
218.Finally, with respect to public procurement, we recommend the establishment of an online bulletin board for the advertisement of challenges which the Government Office for AI and the GovTech Catalyst have identified from across Government and the wider public sector where there could be the potential for innovative tech- and AI-based solutions.
219.The potential impact of AI on the wider economy was one of the most widely discussed and contentious issues of our inquiry. The prospect of significant productivity gains from AI invariably raises the prospect of increased unemployment, although it is equally possible that productivity can grow alongside employment, assuming economic output also increases. The proportion of jobs estimated to be at risk in developed economies such as the UK normally range between 10% and 50%, over the next 10–20 years. The kinds of jobs most at risk are also frequently debated, with some arguing that low-skilled jobs are at far greater risk, while others argue that many white-collar, but relatively repetitive or less creative jobs, might also be at risk.
220.As a number of our witnesses have emphasised, while the current debate stems largely from academic work carried out at the start of this decade, concerns surrounding ‘technological unemployment’ have a long and distinguished history. The 1920s and 1930s saw much public debate on both sides of the Atlantic about the threat of ‘technological unemployment’, a term popularised by John Maynard Keynes. In 1949 Norbert Wiener warned that computerisation, combined with “the valuation of human beings on which our present factory system is based” could usher in “an industrial revolution of unmitigated cruelty”. In the 1960s similar anxieties led to President Johnson establishing the US National Commission on Technology, Automation and Economic Progress in 1964. Professor Edgerton noted the similarities here in the UK, when in 1963 Harold Wilson warned that “computers have reached the point where they command facilities of memory and of judgment far beyond the capacity of any human being or group of human beings who have ever lived”, and speculated that the white collar professions would be particularly hard hit.
221.Contemporary concerns can largely be traced back to Erik Brynjolfsson and Andrew McAfee’s influential 2011 book, Race Against the Machine, which predicted widespread disruption and upheaval as a result of accelerating automation, in part as a consequence of advances in AI. In 2013 Carl Frey and Michael Osborne started a trend for attempting more precise predictions, and by examining the jobs they believed were most susceptible to automation with current or near-future technology, they claimed that around 47% of total US employment was at risk of automation. Jeremy Bowles, applying the same methodology, calculated that 54% of jobs across the EU were similarly threatened.
222.In a 2016 study, Organisation for Economic Co-operation and Development (OECD) economists devised a further methodological innovation, and shifted their attention to focus on how automatable particular tasks within jobs were. While some current jobs may be composed solely of tasks which are completely automatable, therefore rendering the job itself automatable, they concluded that most jobs did not currently fall into this category. Using this methodology, John Hawksworth and Richard Berriman concluded in a 2017 report for PwC that up to 30% of existing UK jobs are at ‘high risk’ of automation by the 2030s. These risks are highest in sectors such as transportation and storage (56%), manufacturing (46%) and wholesale and retail (44%). However, Berriman and Hawksworth believe that due to the additional jobs which are likely to be created through economic growth over this period, the net effect on employment is likely to be neutral.
223.Over the same period, some economists have adopted a more historical approach, by focusing on patterns of automation in the recent past. These include Daron Acemoglu and Pascual Restrepo’s study of the impact of industrial robot usage between 1990 and 2007 on the US labour market, and David Autor’s work on the history and future of workplace automation. While Acemoglu and Restrepo have estimated that areas in the US most exposed to industrial automation in the 1990s and 2000s experienced “large and robust negative effects” on employment and wages, Autor concludes quite the opposite, noting the ways in which automation has often complemented human labour in ways which “increase productivity, raise earnings and augment the demand for labour”.
224.This research has in turn percolated through the policymaking world, influencing a number of important recent reports on the impact of AI and automation on the labour market. The RSA have emphasised the need to consider the impact of AI and automation on the quality and substance of jobs, particularly low-skilled jobs, and the need “to accelerate the adoption of AI and robotics … in a way that delivers automation on our own terms”. Future Advocacy, a London-based think tank, have argued that while the net impact of AI and automation might be relatively neutral, the impact across different regions of the UK is likely to be highly divergent based on their current economic strengths and weaknesses. The IPPR’s report in December 2017 on the subject, drawing heavily on Autor’s approach, argued that automation is likely to transform, rather than eliminate, work, but that policies will be needed to both to accelerate automation in the interests of boosting productivity and wages, and manage the growing inequalities in wealth, income and power which could otherwise arise.
225.Evidence we recieved varied on the likely nature and severity of this impact. One school of thought, generally favoured by businesses and those developing AI systems, believed that the impact would be relatively moderate, or even positive. Tasks, rather than entire jobs, were likely to be automated, and therefore human capacities in many jobs would be augmented, rather than replaced. As the Canadian Institute for Advanced Research’s (CIFAR) evidence explained, “enabling technologies complement and increase the productivity (and wages) of certain types of skills (e.g. laptops for managers and workers specializing in problem-solving, scanners for cashiers). In contrast, replacing technologies conduct tasks previously performed by labour (e.g. assembly tasks, switchboard operation, mail sorting)”. Others argued that even if many types of jobs were entirely automated (AI as a ‘replacing technology’), other jobs would be created in the process, as happened during the nineteenth-century industrial revolution. Very few witnesses provided much detail on what these new jobs might look like, although they may well be impossible to predict.
226.The other broad school of thought came from think tanks and NGOs, and proposed that AI was likely to be far more disruptive to future employment patterns, as many blue- and white-collar jobs might be automated over a very short space of time, hindering the chances for those made redundant to find alternative work. Such witnesses warned that the impact of such change would not be evenly distributed across the country. One witness, with experience in the call centre industry, emphasised the potential scale of the challenge in their own industry:
“Referring again only to the Customer Services industry, in my opinion there will be a reduction in the number of humans required to interact with customers of around 40% by 2020 and that will rise to 70% by 2025. Humans answer around 8.15 billion calls to UK Contact Centres of which there are 7,500 employing just under one million people. HMRC and the Department of Work and Pensions are the largest with 13,000 and 28,000 respectively. That’s 400,000 then 700,000 people who will need to reskill or employ their knowledge in other parts of the business”.
227.However, considerable doubts were raised by later witnesses in our inquiry regarding the methodological soundness of much of the academic literature in this area. Professor Susskind, noting that “there is no evidence from the future”, emphasised that studies which have broken down jobs into their constituent tasks could be misleading, as jobs have frequently been reconstituted around new technologies before, and automated processes do not generally entail simply copying jobs or professions as they existed prior to automation. As a result, Professor Susskind found many of the predictions about job losses to be “entirely unreliable”, and predicting job creation “even harder”, and dismissed the “quasi-science of the major consulting firms” as lacking deep theoretical foundations. Professor Dame Henrietta Moore, Director of the Institute for Global Prosperity, UCL, and Olly Buston, CEO and Founder, Future Advocacy, while still believing in the need for action to counter the possibility of job losses, concurred that many predictions were “evidence light”.
228.In this vein, it is also worth noting that much of the existing literature focuses on the technical potential for automating particular jobs or tasks, which does not necessarily translate into a risk that they will be automated in the real world. Many social and economic factors may influence whether a task is automated or not, beyond the technical potential to do so. What we do know is that employment by sector has changed dramatically over the past century. In 1901, manufacturing accounted for nearly 40% of employment across the country, and agriculture and fishing nearly 10%, but by 2011 these had fallen to 9% and 1% respectively, while the service economy now accounts for more than 80% of all employment (Figure 5). At least part of this change can be attributed to the impact of automation, and we would be remiss if we did not expect considerable change in employment patterns over the next 100 years.
229.A number of recent surveys suggest that the British public are significantly less concerned about automation affecting their own jobs than many experts are. A survey of 2108 adults conducted by YouGov, on behalf of Future Advocacy, between September and October 2017 found that the British public appeared to be comparatively unconcerned about the risks of losing their jobs to automation in the near future (as seen in Figure 6).
Total sample size was 2108 adults. Fieldwork was undertaken between 29th September and 2nd October 2017. The survey was carried out online. The figures have been weighted and are representative of all UK adults (aged 18+).
230.A higher rate of concern was suggested by Demos, in a survey of 1234 adults in October 2017, which found that 35% thought there was “a risk [to their current jobs] from future developments in artificial intelligence and automation”, compared with 53% who believed there was no risk. However, CognitionX also cited a survey conducted by Arm and Northstar, which suggests that internationally concerns may be higher, with 57% of global respondents concerned that AI might become a risk to their jobs.
231.The labour market is changing, and further significant disruption to that market is expected as AI is adopted throughout the economy. As we move into this unknown territory, forecasts of AI’s growing impact—jobs lost, jobs enhanced and new jobs created—are inevitably speculative. There is an urgent need to analyse or assess, on an ongoing basis, the evolution of AI in the UK, and develop policy responses.
232.While the impact of AI on jobs remains highly uncertain, many of our witnesses believed that further Government assistance in terms of adult retraining, reskilling and lifelong learning would be an effective means preparation. In particular, the PHG Foundation argued that this should be focused on “skillsets that arguably cannot easily be displaced by AI such as creativity, effective social interaction, manual dexterity and intelligence”, while Research Councils UK suggested that “in-career re-skilling will become the norm every 10 years”. Dr Ian Morgan and Brian Joyce also observed that, given that “further education has been extensively cut, and course fees at universities are typically excessive for mature students, reducing applications by around 50% over the last 5 years … financial support for those who wanted to retrain would be invaluable”.
233.Future Advocacy highlighted the extent to which re-skilling could mitigate the impact of automation on jobs, providing the example of Accenture, where “17,000 jobs were automated but no-one lost their job, a feat that CEO of financial services Richard Lumb attributed to reskilling”.
234.There were notes of caution as well. Accenture emphasised the importance of ensuring that “those who were left behind by such fast-moving technological developments in the past: minorities, women, working mothers, disabled persons” needed to be included and prioritised in such efforts. Professor Susskind said that consideration would need to be given to existing skillsets, as “the gap between the current skill set of white-collar workers and the toolkit needed for the 2020s is large, and it is not always clear how this gap can actually be bridged”. In blue-collar jobs, he suggested this gap would likely be even bigger, and “truck drivers who are rendered redundant by autonomous vehicles will rarely have the educational background or training to support their simple retraining and redeployment as, say, software engineers”. Future Advocacy highlighted the risks in this particular sector, noting that trials were planned for convoys of semi-automated lorries in the UK by the end of 2018, which posed a risk to the haulage and logistics industry’s 2.2 million employees.
235.For its part, the Government announced in its 2017 Autumn Budget that it would be establishing a National Retraining Scheme, aimed at helping people “re-skill and up-skill as the economy changes, including as a result of automation”. The scheme will be guided by the National Retraining Partnership, which aims to bring together the Government, businesses and workers, through the CBI and the Trades Union Congress (TUC). It will initially focus on “priority skills”, with the first two named areas being digital and construction, funded with an initial investment of £64 million. The scheme will be informed by the results of a £40 million programme to test “innovative approaches to helping adults up-skill and re-skill”, with particular emphasis on the use of AI and other innovative education technologies in online digital skills courses. The Autumn Budget also announced an £8.5 million investment over the next two years in Unionlearn, a subsidiary of the TUC designed to boost learning in the workplace.
236.The UK must be ready for the disruption that AI will have on the way in which we work. We support the Government’s interest in developing adult retraining schemes, as we believe that AI will disrupt a wide range of jobs over the coming decades, and both blue- and white-collar jobs which exist today will be put at risk. It will therefore be important to encourage and support workers as they move into the new jobs and professions we believe will be created as a result of new technologies, including AI. The National Retraining Scheme could play an important role here, and must ensure that the recipients of retraining schemes are representative of the wider population. Industry should assist in the financing of the National Retraining Scheme by matching Government funding. This partnership would help improve the number of people who can access the scheme and better identify the skills required. Such an approach must reflect the lessons learned from the execution of the Apprenticeship Levy.
248 (Andrew de Rozairo)
249 Written evidence from techUK ()
250 Written evidence from Department for Digital, Culture, Media and Sport and Department for Business Energy and Industrial Strategy ()
251 Written evidence from Fujitsu (); Information Technology Industry Council (); Imperial College London () and Arm ()
252 Daron Acemoglu, David Autor, David Dorn, Gordon H. Hanson, and Brendan Price, ‘Return of the Solow Paradox? IT, Productivity, and Employment in US Manufacturing’ in American Economic Review, vol. 104 (2014), pp 394–399: [accessed 14 February 2014]
253 Erik Brynjolfsson, Shinkyu Yang, ‘The Intangible Costs and Benefits of Computer Investments: Evidence from the Financial Markets’, MIT Sloan School of Management (December 1999): [accessed 14 February 2018]
254 Written evidence from The RSA ()
255 (Sarah O’Connor)
257 Written evidence from Center for Data Innovation ()
258 Written evidence from the Institute of Chartered Accountants in England and Wales (ICAEW) ()
259 Written evidence from Sage ()
261 (Kriti Sharma)
262 House of Commons Library, Business Statistics, Briefing Paper, , December 2017
263 Written evidence from CBI ()
264 Written evidence from Research Councils UK ()
265 Written evidence from Vishal Wilde ()
266 , p 39
267 (Matt Hancock MP)
268 Ofcom, ‘New Ofcom rules to boost full-fibre broadband’ (23 February 2018): [accessed 5 March 2018]
269 Written evidence from BSA The Software Alliance ()
270 Written evidence from Professor Richard Susskind ()
271 Written evidence from Doteveryone ()
272 Written evidence from techUK ()
273 Written evidence from Microsoft ()
274 Department for Business, Energy & Industrial Strategy and Department for Digital, Culture, Media and Sport, ‘Recommendations of the review’ (15 October 2017): [accessed 1 February 2018]
275 Written evidence from HM Government ()
276 Kaggle is an online platform, owned by Google, which hosts data science and machine learning competitions to which data and computer scientists compete to develop the best models for handling provided datasets. Rewards are offered (often in terms of cash prizes) for these solutions.
277 (Matt Hancock MP)
278 Written evidence from Royal Society ()
279 National Audit Office, Government’s spending with small and medium-sized enterprises (March 2016): [accessed 12 January 2018]
280 Crown Commercial Service, ‘About us’: [accessed 14 February 2018]
281 Written evidence from SCAMPI Research Consortium, City, University of London ()
282 Written evidence from UK Computing Research Committee ()
283 See Public Sector: Directive 2014/24/EU, Concessions: Directive 2014/23/EU and Utilities: Directive 2014/25/EU
284 Crown Commercial Service, Procurement Policy Note—New Thresholds 2018 (December 2017): [accessed 17 January 2018]
285 The Government’s criteria for designing, building and buying better technology.
286 John Markoff, ‘In 1949, He Imagined an Age of Robots’, The New York Times (20 May 2013): [accessed 30 January 2018]
287 (Professor David Edgerton)
288 Andrew McAfee and Erik Brynjolfsson, Race Against the Machine: How the Digital Revolution is Accelerating Innovation, Driving Productivity, and Irreversibly Transforming Employment and the Economy (Lexington, Massachusetts: Digital Frontier Press, 2011)
289 Carl Frey and Michael Osborne, The future of employment: How susceptible are jobs to computerisation? (17 September 2013): [accessed 1 February 2018]
290 Jeremy Bowles, ‘The computerisation of European jobs’, bruegel.org (24 July 2017): [accessed 1 February 2018]
291 Richard Berriman, John Hawksworth, ‘Will robots steal our jobs? The potential impact of automation on the UK and other major economies’, PwC UK Economic Outlook (March 2017): [accessed 1 February 2018]
292 Daron Acemoglu and Pascual Restrepo, ‘Robots and Jobs: Evidence from US Labor Markets’ NBER Working Paper No w23285 (March 2017): ; David Autor, ‘Why are there still so many jobs? The history and future of workplace automation’ in The Journal of Economic Perspectives, vol. 29, no. 3 (2015), p 5: [accessed 8 March 2019]
293 The RSA, The Age of Automation (September 2017), p 8: [accessed 17 January 2018]
294 Future Advocacy, The impact of AI in UK constituencies: Where will automation hit hardest? (October 2017): [accessed 1 February 2018]
295 IPPR, Managing automation: Employment, inequality and ethics in the digital age (28 December 2017),
p 20: [accessed 17 January 2018]
296 See, for example, written evidence from the RSA (); Dr Ian Morgan and Brian Joyce () and euRobotics Topics Group on ‘Ethical, Legal and Socio-economic issues’ ()
297 See, for example, written evidence from Fujitsu (); Information Technology Industry Council ITI (); Imperial College London () and Arm ()
298 Written evidence from CIFAR ()
299 See, for example, written evidence from Braintree (); CBI (); CIFAR (); Fujitsu (); Innovate UK (); Microsoft (); BSA The Software Alliance (); The RSA () and euRobotics Topics Group on ‘Ethical, Legal and Socio-economic issues’ ()
300 See, for example, written evidence from Charities Aid Foundation (); The Economic Singularity Supper Club (); The Knowledge, Skills and Experience Foundation (); Professor Maja Pantic (); Dr Mike Lynch (); Warwick Business School, University of Warwick (); Deep Science Ventures () and Dr Aysegul Bugra, Matthew Channon, Dr Ozlem Gurses, Dr Antonios Kouroutakis and Dr Valentina Rita Scotti ()
301 Written evidence from Contact Centre Systems Ltd. ()
302 (Professor Richard Susskind)
303 (Professor Henrietta Moore and Olly Buston)
304 Data for 1901 to 1911 covers Great Britain, while data for 1921 to 2011 covers only England and Wales. There is no census data for 1941, and the ONS did not include data for 1971 due to the difficulties of creating a consistent industrial grouping.
305 Written evidence from Future Advocacy ()
306 Demos, ‘Public views on technology futures’ (29 November 2017): [accessed 1 February 2018]
307 Written evidence from CognitionX ()
308 See for example Dr Toby Walsh (); PHG Foundation (); Amnesty International (); Research Councils UK (); Professor Richard Susskind (); CognitionX () and IBM ()
309 Written evidence from PHG Foundation ()
310 Written evidence from Research Councils UK ()
311 Written evidence from Dr Ian Morgan and Brian Joyce ()
312 Written evidence from Future Advocacy ()
313 Written evidence from Accenture UK Limited ()
314 Written evidence from Professor Richard Susskind ()
315 Written evidence from Future Advocacy ()
316 , p 41
317 , p 11
318 , p 117
319 , p 47