Unpacking the UK’s approach to AI regulation

The recent 2^nd reading of Lord Holmes’ Artificial Intelligence (Regulation) Bill in the House of Lords provides a fascinating insight into the myriad of issues and themes surrounding regulation of artificial intelligence (AI) in the UK. The debate underscored the complexity and multifaceted nature of AI, revealing a landscape where innovation, ethics, legal obligations, and societal impacts intersect. Referring to the scale of the challenge, Viscount Chandos said “there can have been few Private Members’ Bills that have set out to address such towering issues as this bill”. This article seeks to comment on some of the key themes and issues raised during this crucial parliamentary discourse.

Ethical and societal implications of AI decisions

One of the central concerns highlighted was the ethical and societal implications of AI decisions. As Lord Holmes, referencing the Policy briefing: Regulating AI in the UK, poignantly asked: “Where do you go if on the wrong side of an AI decision?”. This question not only underscores the potential for harm and injustice through the deployment of AI, but also the urgent need for clear and accessible redress mechanisms. The Bill envisages that the “AI Authority” be a “co-ordinator of regulators” sitting above and looking across existing regulators to assess their competency to address the opportunities and challenges presented by AI. One of the proposed roles of the regulatory “AI Authority” would be to carry out a gap analysis of regulatory responsibility in respect of AI. This appears to be a much-needed exercise as the current regulatory regime lacks clarity. For example, as mentioned in the debate, AI is making a huge contribution in education, but which is the regulator? The DfE? Ofsted? Ofqual? Which does one turn to if AI is being used to mark exam papers and gets it wrong?

Divergent views on regulation

The sentiment amongst those that spoke in the debate appeared to lament the Government’s lack of initiative with regard to introducing regulation in this area. While the majority supported the establishment of a regulatory body, this was not unanimous. Concern over the establishment of an “AI Authority” was voiced by Baroness Stowell asserting that regulation might create barriers for entry and so hinder an open market for AI innovation, which in turn would lead to greater concentration of power among large technology companies. This divergence in views underscores the challenge in balancing the need for good governance and the promotion of a vibrant AI sector in the UK economy that is capable of competing globally and enabling other sectors of the economy to do so too.

Intellectual property and data use

Staying on the competitive theme, it was mentioned that artists have expressed fear that human-made artworks will have no distinctive or unique value in the marketplace in which they operate, and that AI may thereby render them obsolete. A case was made by Lord Freyberg for the need to support creative artists under threat from AI by protecting their copyright and providing fair compensation where their work is used to train AI. The Bill contains provisions requiring reporting of the use of third-party data to the AI Authority and shows a legislative intent to prevent AI technologies infringing the rights of copyright holders and data subjects without informed consent.

Transparency and public engagement

The calls for high transparency, particularly through labelling when AI is used in the provision of services, reflects a broader demand for informed consent and awareness of how data is handled, and decisions are made. Baroness Finlay of Llandaff, quoted the Dutch proverb of “Trust arrives on foot and leaves on horseback” and if public trust in AI is not to be ephemeral any regulatory approach will need to ensure that the current foundations being laid for AI support a safe future. To that end the Bill makes specific provision for public engagement, mandating that the AI Authority “implement a programme for meaningful, long-term public engagement about the opportunities and risks represented by AI”. How that may manifest is unclear, but it demonstrates that building and maintaining public confidence in this fast-evolving technology will be key.

Sandboxes and AI Responsible Officers

The debate also touched on the use of regulatory “sandboxes”, continuing the approach taken with the fintech regulatory sandbox in 2016, as a means to foster safe AI experimentation within a controlled environment. This approach could balance the need for innovation with the imperative to protect public interests. However, the sandbox approach has not been without criticism as it could be seen as the regulator providing a business with some kind of stamp of approval or free marketing given the value of the PR that involvement in a sandbox may bring.

Additionally, the proposal for an “AI Responsible Officer” for businesses underscores the growing recognition of the need for accountability and appropriate oversight in the deployment of AI technologies, but would be a significant change for the governance of UK business. The objectives of the duties of such an officer as set out in the Bill are worthy, how would those duties be discharged and what liability would that officer have and to whom?

Conclusion

This debate over the regulatory approach proposed by the Bill has highlighted the intricate interplay between innovation, ethics, and regulation in the AI landscape. The issues will no doubt be explored in more detail as the Bill goes through its committee stage and the government’s current position that it is too soon to legislate effectively and potentially be counterproductive to do so, may come under pressure. This stands in contrast to the EU’s approach, which has already implemented wide-ranging legislation to regulate AI. While consumers, employers, workers, and civil society are in the foothills of an AI-driven future, a clear direction on regulation needs to develop to give clarity, certainty, and consistency to help guide their ascents.