Artificial Intelligence (or machine learning as some experts would prefer we called it) has been a source of both excitement and concern for some time. The recent international summit, hosted by the UK at Bletchley Park, focussed largely on the concern. It brought together the leadership of many of the leading companies developing AI and world leaders who will have to decide what constraints should be put on this transformative technology’s expansion into multiple areas of our lives. The Prime Minister deserves huge credit for his determination to ensure these questions are addressed at the most senior level, and for his mastery of the subject at a crucial time. The summit produced important commitments on the testing of AI models, but it neither represented the last word on the subject of AI regulation nor claimed to. That is why I find it slightly odd that this week’s King’s Speech contained no specific commitment to regulation on that subject.
The King’s Speech is not just a big moment for our new Monarch, but also for the Government. It will almost certainly be the last before a General Election and, among the waves of events to which Governments have to react, is a significant chance to chart their own course. It should be a mix of the urgent and the important, and AI is surely both. I understand the Government’s wish to avoid stifling innovation by encouraging good behaviour among AI developers rather than compelling it and, as I say, the Bletchley Park summit showed that progress can be made in doing so, but I am skeptical that the establishment of a clearer regulatory framework for AI can be avoided for long. I have some experience of legislating to regulate technology and I would say there are 2 lessons to be learned from the process which led to the Online Safety Act. The first is that balanced and effective legislation can provide the clarity which assists, not stifles innovation, especially in an area of technology where innovators can see that some regulation is inevitable at some point. I would say this applies to AI. The second is that it is much harder and much less desirable to retrofit regulation onto a developed technology than to set out the rules of the road while the technology is still developing and can take account of those regulatory principles. With AI, the speed of development is such that we have no time to waste. I also know that legislation in these areas takes a long time to get right, and that Parliament needs to be involved early. For these reasons, though I welcome much of what His Majesty set out of his Government’s legislative intentions, I would have liked to have heard more about AI regulation.
