Lord Falconer of Thoroton (Lab)

I am grateful to the noble Baroness, Lady Coffey, for raising artificial intelligence. There was, broadly, a consensus around the Committee, which the noble Baroness supported, that the amendment is much too blunt, but as she said, fairly, it gives us an opportunity to talk about AI. I will also pick up the right reverend Prelate the Bishop of Hereford’s contribution; he rightly said, as has been echoed around the Committee, that there have been huge benefits for patients from AI. I think four concerns were raised during the debate. The first was: will AI affect decision-making? I think the underlying point there is that we do not want machines to make the decisions that are referred to in the Bill; we want human beings to make them. In particular, the decisions I have in mind are the decision of the first doctor, the decision of the second doctor, the decision of the panel, and the decision of the doctor, at the point that the assistance is being given, that the conditions are still satisfied. Everybody around the Chamber wants that to be decided by a doctor or a panel, depending on which it is, and I completely and unreservedly endorse and accept that. Does that need to be made even clearer in the Bill? I will consider it, but I do not think that it does. The acid test for me is that if you fail to comply with your obligations as a doctor or as a panel, you can go to prison for up to five years. It is very difficult to imagine how you could put a machine in prison, so it is pretty clear that these decisions must be made by a human being. For my part and for everybody who supports the Bill, that must remain the position. The second concern is advertising, which the noble Baroness, Lady Berridge, referred to. She is absolutely right. I have made it clear that I will bring forward amendments. Those amendments, which are almost finally drafted, make provision specifically in relation to digital advertising—they do not specifically refer to AI, but we need to address that in the advertising provision. I will lay those amendments so that the House can consider them. The third concern is slightly generalised, which is that AI is very persuasive, particularly in persuading people to do things that they do not necessarily want to do. The first thing on that is that there is a wider societal requirement to address the pervasive impacts of AI in a whole range of things. We should all try to contribute to that. More focused on this is the question of the safeguards in the Bill, because they then become incredibly important. In particular, the safeguards require that there is doctor-to-patient discussion in relation to the decision for that patient, and they are specifically required in the preliminary conversation, the first conversation and the second conversation. It is those safeguards that one must see as the antidote to the persuasive aspect of AI, but I completely accept what people said on that. The fourth issue, which was touched on very briefly, was the operation of devices. That, I think, referred to the fact that quite a number of medical devices can be operated by, for example, the blink of an eye or something quite minor. Again, that needs to be properly safeguarded. Those may not necessarily be AI problems but problems with other sorts of developments in technology. I thank the noble Baroness, Lady Coffey, for raising this. We need to consider all the points she made. At the moment, apart from the advertising amendment, which I will bring forward, I am not sure that it requires amendment to the Bill.