The Select Committee on Artificial Intelligence held a roundtable meeting, hosted by techUK, to discuss the opportunities and challenges for small and medium-sized enterprises (SMEs) who are developing or using artificial intelligence.
Five members of the Committee were in attendance, as was Dr Mateja Jamnik, Specialist Adviser to the Committee.
The session was attended by representatives from: Accenture (UK) Ltd, Access Partnership, Adarga Ltd, Advanced, Baker & McKenzie LLP, Blue Prism Group Plc, Bristows LLP, BSI Group, Cisco Systems Ltd, Cloudera (UK) Ltd, CMS Cameron McKenna Nabarro Olswang LLP, DeepMind, DigitalGenius, Emeiatec, Five AI, FTI Consulting, Gavin Jones Consulting Ltd, Google, IBM United Kingdom Ltd, iPLATO Healthcare Ltd, Kensai, Nominet, Oracle Corporation UK Ltd, PI Ltd, Pivigo Ltd, SAP (UK) Ltd, SVGC Ltd, Unilink Software Ltd, Vodafone Ltd, Your.MD and techUK.
The group began by discussing the question: “Why has your business decided to develop and/or deploy artificial intelligence? What benefits does it offer your business and your customers?” The increasing availability of data was frequently mentioned, as was the inability of many organisations to process and understand this data, which was in turn hampering efforts to boost productivity. AI was thought to present a solution to this problem. It was also noted that in many cases, AI was being introduced ‘via the back door’, with companies signing up to products and services which happened to make use of AI, rather than explicitly seeking out ways in which they could utilise it.
The discussion then focused on the question: “What barriers have you encountered in trying to develop and/or deploy artificial intelligence?” Many of the businesses present had views on the accessibility of funding, and it was generally agreed that securing funding to scale up companies from around 10 to 100 employees could be challenging in the UK, as UK investors were considered more risk averse than their US counterparts.
The difficulty in attracting skilled AI developers was also highlighted, as there was a general shortage in the UK and elsewhere, and those that were to be found attracted high salary premiums. Other points which were made included the perceived inflexibility of the apprenticeship levy, and the number of businesses which were still not using today’s technology, and were therefore not in a position to adopt AI in the near future.
The group was asked: “Is the UK a good environment for start-ups focusing on developing or deploying artificial intelligence? What could help improve the environment?” They responded by highlighting the role of the catapults, particularly the Digital Catapult, which was an important way in which start-ups could access equipment and advice. It was observed that more expertise was needed within companies in terms of what AI could do, so that clients could lead in developing AI-orientated solutions to real-world problems, rather than the current situation, where start-ups sometimes attempted to solve problems which did not actually exist.
The group then discussed the question: “What ethical considerations does your business make when considering the development or use of an artificial intelligence system?” It was emphasised that the public needed confidence in AI systems, and the decisions they made, if they were going to accept them. New frameworks were needed for explainability, and explainability also needed to be designed into systems from the start, as this capacity was difficult or impossible to retrofit into existing systems. New mechanisms would be needed for tracing liability as well.
More broadly, many at the event felt that ethical guidance, and a code of ethical practice in AI, was needed in addition to more clarity on the implications of new legislation like the GDPR. When the ethics of valuing data were brought up, it was emphasised that further guidance on appropriate sharing, and reciprocal arrangements, would be of use. They noted that many public organisations seemed not to know what kinds of data they could and could not share, and that further clarity in this area would reap dividends.
When asked what they would like to see from a Government AI Sector Deal, a number of points were raised. Many attendees thought that government procurement could be a major boost to AI start-ups in the UK if government departments could be encouraged to look to UK companies first. Currently, EU regulations limited what could be done in this area, but post-Brexit, some attendees felt that there would be opportunities to reassess this.
Most attendees also believed that the government needed to invite AI start-ups and SMEs to Whitehall more frequently than they currently do. Immigration restrictions were considered an issue, as many of the most skilled AI developers in the UK today had come from abroad, and this flow needed to be maintained. Many attendees were sharply critical of Innovate UK, and believed it was not adequately serving the AI development sector. Several attendees also discussed the need for greater co-operation between industry and academia, especially from universities outside of the more developed Russell Group, and it was felt that the Government could play a more active role here.
Despite these particular issues, overall there was a view that the Industrial Strategy was broadly a step in the right direction—what was needed now was action on these policies, within a reasonably rapid timeframe, and an understanding of who, exactly, would be held accountable for their execution.
649 Lord Clement-Jones (Chairman), Baroness Grender, Lord Puttnam, Baroness Rock and Lord St John of Bletso.