Terminally Ill Adults (End of Life) Bill

Debate between Baroness Lawlor and Lord Wolfson of Tredegar

Committee stage
Friday 30th January 2026 Lords Chamber

[Baroness Lawlor (Con)]

My Lords, to add to what has been said, AI is based on large language models, which involve big datasets. I ask your Lordships to consider whether such large datasets, based on assessing a snippet of data to assist diagnosis, are a good way of assessing individual patients. They were not designed to assess individual patients. Every doctor will tell you that each individual case is different, and that diagnosis can vary. I am very grateful to the noble Lord, Lord Stevens, for sharing the results of the 98,000 cases that were assessed for accuracy. Therefore, I am not sure that it is a suitable tool to assess and diagnose individual cases.

[Lord Wolfson of Tredegar (Con)]

My Lords, I approach this amendment in a largely exploratory spirit, because it raises an interesting and increasingly relevant question about the role artificial intelligence might play in the operation of legislation of this kind.

Much of what the Bill requires involves assessments of different natures. We have discussed a number of these at length in Committee, such as questions of capacity and voluntariness, and the presence or absence of coercion. These are judgments that are normally understood as being made by human decision-makers, who draw on their professional experience and have a direct interaction with the individual concerned. At the same time, this is 2026, and we are all aware that artificial intelligence is becoming more prevalent across the Civil Service and even in government departments. Sometimes, in relation to the latter, I am reminded of the Imelda Marcos line, “Better nouveau riche than no riche at all”—and perhaps better artificial intelligence than no intelligence at all. But the concern is that the use of AI is often incremental and not the result of a conscious policy decision or choice. In response to my noble friend, I am not aware of an entire judgment put together by artificial intelligence, but I am aware that sometimes, judgments have been found to dip into artificial intelligence to assist.