AI in the UK (Liaison Committee Report)

Lord Evans of Weardale Excerpts

Wednesday 25th May 2022 Grand Committee

[Lord Evans of Weardale (CB)]

My Lords, I draw attention to my entry in the register of interests as an adviser to Luminance Technologies Ltd and to Darktrace plc, both of which use AI to solve business problems.

I welcome the opportunity to follow up the excellent 2018 report from the Select Committee on Artificial Intelligence. In 2020 the Committee on Standards in Public Life, which I chair, published a report, Artificial Intelligence and Public Standards. We benefited considerably from the work that had gone into the earlier report and from the advice and encouragement of the noble Lord, Lord Clement-Jones, for which I am very grateful.

It is most important that there should be a wide-ranging and well-informed public debate on the development and deployment of AI. It has the potential to bring enormous public benefits but it comes with potential risks. Media commentary on this subject demonstrates that by swinging wildly between boosterism on the one hand and tales of the apocalypse on the other. Balanced and well-informed debate is essential if we are to navigate the future successfully.

The UK remains well-positioned to contribute to and benefit from the development of AI. I have been impressed by the quality of the work done in government in some areas on these underlying ethical challenges. A good example was the publication last year of GCHQ’s AI and data ethics framework—a sign of a forward-looking and reflective approach to ethical challenges, in a part of government that a generation ago would have remained hidden from public view.

The view of my committee was that there was no reason in principle why AI should not both increase the efficiency of the public service and help to maintain high public standards, but in order to do so it had to manage the risks effectively and ensure that proper regulation was in place, otherwise public trust could be undermined and, consequently, the potential benefits of AI to public service would not be realised. The Liaison Committee report gives me some encouragement about the Government’s direction of travel on this, but the pace of change will not slow and continuing attention will be required to keep the policy up to date.

Specifically, I welcome The Roadmap to an Effective AI Assurance Ecosystem by the CDEI, which seems to me, admittedly as an interested layman rather than a technologist, to provide realistic and nuanced guidance on assurance in this area—and it is one where effective independent assurance will be essential. I therefore ask the Minister how confident he is that this guidance will reach and influence those offering assurance services to the users of AI. I welcome the consultation by DCMS on potential reforms to the data protection framework, which may need to be adjusted as advances in technology create novel challenges. I look forward to seeing the outcome of the consultation before too long.

The Government’s AI strategy suggests that further consideration will be given to the shape of regulation of AI and is to be published later this year, specifically considering whether we are better to have a more centralised regulatory model or one that continues to place the responsibility for AI regulation on the sectoral regulators. Our report concluded that a dispersed vertical model was likely in most areas to be preferable, since AI was likely to become embedded in all areas of the economy in due course and needed to be considered as part of the normal operating model of specific industries and sectors. I remain of that view but look forward to seeing the Government’s proposals on the issue in due course.

One area where we felt that improvement was needed was in using public procurement as a policy lever in respect of AI. The public sector is an increasingly important buyer of AI-related services and products. There is the potential to use that spending power to encourage the industry to develop capabilities that make AI-assisted decision-making more explicable, which is sometimes a problem at present. The evidence that we received suggested that that was not being used by government, at least as recently as 2020. I am not sure that we are doing this as well as we should and would therefore welcome the Minister’s observations on this point.