Baroness Fox of Buckley (Non-Afl)

My Lords, the Government’s 10-year health plan for England seeks to “make the NHS the most AI-enabled health system in the world”. Like others, I think that is an incredibly exciting prospect. I do not want it to be dystopian. I think that the right reverend Prelate the Bishop of Hereford makes an important point in warning us against going completely over the top. I think it is important that this amendment has been tabled, because it makes us think about what the possible problems are, which have been well expressed by others. Despite my excitement about what AI might do, even in terms of treatments—there are wonderful possibilities in terms of helping people to walk, what is happening with the brain, and so on—we do not want to be naive. The question for the noble and learned Lord, Lord Falconer, is: as the NHS digitises and doctors become increasingly reliant on AI for notes and diagnostics, given that the diagnosis is so important in a life-or-death situation in this instance, how can we ensure that a time-poor doctor does not use AI as an assessment tool or a shortcut? We would be naive to imagine that that does not happen elsewhere; we would only have to think of politics. People now use AI to avoid doing research, in a wide range of instances, and I do not want that to be translated over. As for the patients, algorithms are supremely impressive and can take things that have happened on Facebook or TikTok, from when you have been on a Teams meeting or Zoom—all sorts of indications—and detect chronic illness conversations. The algorithms can then push pro-assisted dying content such as the Switzerland adverts or positive end-of-life options. Interestingly, when discussing banning social media for under-16s, which I completely disapprove of, or bringing in the Online Safety Act, which I argued against, everybody kept saying, “Algorithms, oh my goodness, they can do all these things”. We should consider not that chatbots are malevolent but that AI tends to agree with people via the algorithms; to quote the title of a piece in Psychology Today, “When Everyone Has a Yes-Man in Their Pocket”. If you say that you are interested in something, they will just say, “Yes, here are your options”. That is something to be concerned about, and it will come up when we discuss advertising. I finish with that BBC story from August of a Californian couple suing OpenAI over the death of their teenage son. They allege that ChatGPT encouraged him to take his own life, and they have produced the chat logs between Adam, who died last April, and ChatGPT that show him explaining his suicidal thoughts. They argue that the programme validated his most harmful and self-destructive thoughts. I am just saying that AI is a wonderful, man-made solution to many problems, but if we pass a Bill such as this without considering the potential negative possible outcomes, we would be being irresponsible.