Artificial Intelligence: Impact on Human Relationships and Society

Lord Kamall Excerpts

Friday 5th June 2026 Lords Chamber

[Lord Kamall (Con)]

My Lords, I too thank the most reverend Primate for securing this important debate. While the speech outlined some of the more existential questions, I will try to build on the most reverend Primate’s comments on healthcare. It is also a pleasure to follow the noble Baroness, Lady Stuart, and I want to build on some of the comments she made about judgment.

In considering the demographic challenge of an ageing population, the increased demand on health and social care and the pressure on the public purse, the recently departed Secretary of State for Health and Social Care suggested that, instead of just spending more money and recruiting more people, the health system should use more AI. When we speak about AI, much of what we call AI today is really machine learning, trained on huge amounts of data to reveal hidden patterns, make predictions and learn over time.

We have the wonderful situation where, when a patient has an ocular scan, not only can the optometrist check the condition of the eye but, thanks to machine learning trained on huge medical data sets, they are able to identify whether patients also have other conditions, such as high blood pressure. So you can see why many are enthusiastic about the potential of AI in medical diagnosis but also in helping clinicians to make more informed decisions. However, taking that a step further and automating decision-making should give us pause for thought. Going even further, when we allow AI systems to rewrite their own algorithms, some see that as a step too far, with fears of machines enslaving humans and ruling the world.

Some of that AI is actually with us now. In a documentary about the use of AI by Ocado—you can see what an interesting guy I am when I watch documentaries like that—the manager explained that, while humans wrote the original algorithm, the system itself rewrites the algorithm to improve the efficiency of preparing the crates for delivery. He admitted that he no longer understood or knew what the algorithm was. For some, that will sound scary, but so far no Ocado robotic pickers have broken out of the warehouse, rampaged through the nearest town and left a trail of destruction.

With regard to medical applications in other areas, especially the military, there are concerns about fully automated processes, as a number of noble Lords have said, but even here it is not always clear cut. Consider AI-driven missile systems. While some have a human in the loop—that is, the AI identifies a legitimate target for a missile strike but a human operator still has to press the button—what happens when, effectively, the human operator says, “Actually, that system gets it right most of the time”, and just automatically presses that button? The same thing happens when we click the button for cookies—we just automatically click. That is a process of self-automation. Now imagine that in the healthcare system.

Interestingly, as an aside, when an AI algorithm was blamed for the recent missile strike that tragically murdered 180 schoolchildren in Iran, as mentioned by my noble friend Lady Helic, it turned out that the US military had not updated the data on that building, demonstrating that AI is not only only as good as the algorithm but only as good as the data it is trained on.

The other concern is that, while we see more use of AI in tech and commerce, the same systems may not always work in healthcare. I shall illustrate with a couple of examples. A few years ago, I arrived at an airport and scanned my boarding pass to get to the gate, but when I went to board the plane, the system was not working. I asked easyJet staff about it, and I was gaslit by many; over the next hour I had meeting after frustrating meeting, trying to find someone who would help me. The next day they admitted to me that a flag had gone off in the system—as if that explained everything.

Another example is when I applied for a Monzo bank account. I got a message saying, “We will process your application within 48 hours”, but a week later I had heard nothing. I chased them up, and a week later I got a message saying, “We’ve decided to reject your application, and by law we don’t have to tell you why”. The point here is that, while those companies can get away with that because there is competition and choice, imagine that happening in healthcare. You turn up for your operation, to be told, “I’m sorry, we don’t have to tell you why you’ve been declined for your operation, but you can’t get in”. So, while we should be excited about the huge potential of AI for medical research and diagnosis, when it comes to combined AI and automation for delivering health and care services, of course let us continue to innovate, but let us do so with caution and humanity.