1.In his seminal paper, Computing Machinery and Intelligence, Professor Alan Turing began by posing a deceptively simple question: “Can machines think?” The question, in one guise or another, has been a source of inspiration for modern literature, drama and art, as well as being a subject of continued scientific endeavour. Yet Turing quickly dismissed it as too ambiguous, instead reformulating the question and describing his ‘Imitation Game’; a test he proposed as a means to establish whether a machine could act indistinguishably from a human. In concluding his paper, he hoped that “that machines [would] eventually compete with men in all purely intellectual fields”, perhaps beginning with “the playing of chess”.
2.In the 66 years since Turing published his landmark paper, the development of what we now term ‘artificial intelligence’ has gone through periods of optimism and progress, only to be followed by setbacks. While machines still do not compete with humans “in all purely intellectual fields”—as Turing put it—artificially intelligent machines have made extraordinary progress in the area he initially singled out: playing, and winning, at board games.
3.Early this year, for example, Google DeepMind’s AlphaGo—an artificially intelligent computer programme—won a five-match series of the ancient Chinese board game ‘Go’ against the reigning world champion, Lee Sedol. Go was “widely viewed as an unsolved ‘grand challenge’ for artificial intelligence” and AlphaGo’s success marked a watershed moment in its ongoing development. Significant progress, however, has been made across the field in recent years, linked to the rise in processing power, the profusion of data and the development of techniques such as ‘deep learning’. Much of that progress—such as improved automated voice recognition software, predictive text keyboards on smart phones and autonomous vehicles—has been driven by UK-based technology start-ups, founded by graduates of UK universities, as well as universities themselves.
4.There is no single, agreed definition of artificial intelligence (AI), though there is a tendency to describe AI by contrasting it with human intelligence and stressing that AI does not appear ‘in nature’. At present, the capacity of ‘AI machines’ is narrow and specific; they can complete what Margaret Boden, Professor of Cognitive Science at the University of Sussex, has described as “specialised tricks”. For example, Google DeepMind’s AlphaGo system cannot “for the moment do anything besides play Go”. Thus, as it currently stands, AI can be loosely thought of as:
a set of statistical tools and algorithms that combine to form, in part, intelligent software that specializes in a single area or task. This type of software is an evolving assemblage of technologies that enable computers to simulate elements of human behaviour such as learning, reasoning and classification.
5.Progress has recently been made in ‘machine learning’—a “way of achieving a degree of AI”. Machine learning involves building algorithms that can learn specific concepts for themselves, without being explicitly programmed. This, in turn, relies on those algorithms processing vast quantities of ‘training data’ in order to learn to identify a statistical rule that correlates inputs with the correct outputs. This type of ‘narrow’ AI is already found in aspects of daily life, from using voice recognition software on a smart phone, to filtering spam out of an email inbox.
6.Machines have also become more adept at translating one language into another, though they do not ‘understand’ language in the same way as a human. They struggle to cope, for example, with syntax and do not comprehend the meaning or implications of the language they are translating. The ‘general’ artificial intelligence—akin to human intelligence—that this would require has not yet been developed. There is continuing debate about when such general artificial intelligence might be achieved, as well as whether it is even possible. According to Professor Stephen Hawking and others, while it might be “tempting to dismiss the notion of highly intelligent machines as mere science fiction [...] this would be a mistake, and potentially our worst mistake ever”.
7.Robotics—machines that are “capable of carrying out a series of actions on behalf of humans”—is a different topic to AI. Robots can (and, for the most part, do) operate without possessing any artificial intelligence. It is anticipated, however, that this will gradually change over time, with robots becoming the ‘hardware’ that use, for example, machine learning algorithms, to perform a manual or cognitive task. AI and robotics will, therefore, have an important degree of interdependency. As one commentator explained, “there is no AI without robotics […] intelligence and embodiment are tightly coupled issues”. For these reasons, our inquiry has considered robotics and AI together.
8.Both robotics and artificial intelligence are complex, and potentially transformative, emerging technologies in which the UK is playing a leading role. Yet it is often difficult to predict with any accuracy how technologies will unfold and evolve. The implications of new technologies tend, therefore, to be examined and understood by policymakers too late in the day to engage with them in any significant way. As a result, technology “is sometimes presented to us as if [it] is on a relentless track in a particular direction and we have no power to move it either way”. We decided to examine robotics and AI after the Government was unable to produce a short statement outlining the evidence underpinning its policy on AI, which we requested as part of our ‘evidence check’ work.
9.By undertaking our inquiry now, we hope that it will be soon enough to be productive and late enough to be relevant. Indeed, the announcement in the Queen’s Speech of the Modern Transport Bill—with its aim to “put the UK at the forefront of autonomous and driverless vehicles ownership and use”—was a stark reminder that advances in robotics and AI are starting to make their way into the mainstream. Other countries are also beginning to look at the wider issues raised by AI. During the course of our inquiry, for example, the White House Office of Science and Technology Policy ran a series of workshops on the implications of AI and launched its own review—Preparing for the Future of Artificial Intelligence.
10.Our inquiry took a broad focus and examined robotics and AI in the round: identifying their potential value and capabilities, as well as examining prospective problems, and adverse consequences, that may require prevention, mitigation and governance. We launched our inquiry in March 2016 and sought written submissions addressing the following points:
a)The implications of robotics and artificial intelligence on the future UK workforce and job market, and the Government’s preparation for the shift in the UK skills base and training that this may require.
b)The extent to which social and economic opportunities provided by emerging autonomous systems and artificial intelligence technologies are being exploited to deliver benefits to the UK.
c)The extent to which the funding, research and innovation landscape facilitates the UK maintaining a position at the forefront of these technologies, and what measures the Government should take to assist further in these areas.
d)The social, legal and ethical issues raised by developments in robotics and artificial intelligence technologies, and how they should be addressed.
11.We received 67 written submissions and took oral evidence from 12 witnesses including:
We also visited Google DeepMind in King’s Cross, London (see Annex). We would like to thank everyone who contributed to this inquiry. In Chapter 2 we look at the economic and social implications of robotics and AI, particularly in the context of the future of work, employment and skills. Chapter 3 focuses on the ethical and legal issues that may be raised, and what governance frameworks might be required, while Chapter 4 examines the research, funding and innovation landscape for robotics and AI.
1 A M Turing, Computing Machinery and Intelligence, Mind, vol 49 (1950), pp 433–460.
2 , The Guardian, 15 March 2016. Go is an abstract strategy board game for two players, in which the aim is to surround more territory than the opponent. Despite having relatively simple rules, Go is considered more complex than chess, having both a larger board with more scope for play and longer games, and, on average more alternatives to consider per move.
3 Google DeepMind () para 1.4.
4 , The Economist, 9 May 2015
5 , Business Insider UK, 5 January 2016; , Digital Catapult Centre, 20 March 2016
6 Gary Lea, , The Conversation, 2 September 2015
7 Professor Margaret Boden, Human-level AI: Is it Looming or Illusory?, lecture at The Centre for the Study of Existential Risk, Cambridge, 19 June 2015
8 Google DeepMind () para 1.5
9 Transpolitica () para 1.4
10 The Royal Society, ’, last accessed 31 August 2016
11 Professor Margaret Boden, Human-level AI: Is it Looming or Illusory?, lecture at The Centre for the Study of Existential Risk, Cambridge, 19 June 2015; see also Professor Tony J Prescott () paras 6 & 9
12 See, for example, The Royal Society (); Professor Huw Price () para 7
13 Stephen Hawking, Stuart Russell, Max Tegmark, Frank Wilczek, , The Huffington Post, 19 June 2014
14 Innovate UK () para 6
15 See, for example, RACE, UK Atomic Energy Authority ()
16 Jean-Christophe Baillie, “I”, March 2016
17 Jack Stilgoe, Science, ethics and shared space, The Guardian, 1 May 2013
18 Q64 [Richard Moyes]
19 For further information on Evidence Check, see Science and Technology Committee, Sixth Report of Session 2016–17, Evidence Check: Smart Metering of Electricity and Gas, HC 161. See also Department for Business, Energy and Industrial Strategy ()
20 The Queen’s Speech 2016, Background Notes, last accessed 3 August 2016 at:
21 White House Office of Science and Technology Policy, , May 2016
5 October 2016