Artificial Intelligence: Impact on Human Relationships and Society

Baroness Teather Excerpts

Friday 5th June 2026 Lords Chamber

[Baroness Teather (LD)]

My Lords, it is a privilege to have the opportunity to take part in this debate. I thank the most reverend Primate for making space for it and for her incredibly helpful and thoughtful analysis of the challenge that we face. As many have noted, this takes place just one week after His Holiness the Pope issued his encyclical Magnifica Humanitas, urging us to reclaim a vision of what it is to be fully human and to put that at the heart of the criteria we use to guide technological progress on AI. Faith leaders, reading the signs of the times, are playing a vital role here: urging us all to consider the wider philosophical, ethical and moral questions around this advancing technology, and how we want to shape it to support human flourishing, not just how we might fix technocratic problems that we can pick off one at a time and compartmentalise.

The scale of AI’s potential impact, on all areas of life for good and ill, requires this breadth of thinking. My first question to the Minister is: where is this cross-cutting, ethical thinking taking place in government? Who is holding the bigger questions about the potential impact of AI on the shape of society? How might the public engage in this wider conversation? I am particularly thinking about those who will be most affected, such as young people, about whom many noble Lords have already spoken.

Also, where are the negotiations at an international level taking place? What are the shared values that we are seeking to influence? I recognise the challenges of that, but if we could at least articulate what we want to do, it would be transparent for all to see.

I will briefly touch on a couple of themes about how AI, without that shaping, could threaten our ability to think and make judgments about the world. The first is about its potential to revolutionise access to research and information. While AI is already a game-changer for scientific and medical research, as so many have already said, this tool that promised to aid the pursuit of truth is also threatening our ability to discern fact from fiction. Already, it is being used to generate and promulgate fake information, which gets presented with great certainty. A report by Demos last week warned that one in three adults reported seeing political deepfakes generated by AI immediately before the recent local and devolved elections. The research also suggested that some are being exposed to much higher levels of deepfakes than others. The coupling of AI generated misinformation with personalised targeting of content through social media feeds is a toxic mix undermining democracy and driving division. It is also a force that is often hidden from view, because we have no idea what anyone else is seeing. This is what threatens our common understanding of fact, which the right reverend Prelate the Bishop of Leicester discussed.

The second theme I will briefly touch on is automated decision-making, which is finding its way into systems governing access to benefits, housing, immigration and more. Without a human in the loop—a human who is continuing to think and to hold accountability—it has potential to create unimaginable levels of injustice, which become progressively more difficult to see, challenge and unpick. We have been here before, of course: more than 900 postmasters and postmistresses were accused of theft on the basis of the Horizon system. There was also the TOEIC scandal, where voice software led to 2,500 students forcibly removed from the UK. Overreliance on computer systems erodes human capacity to ask the obvious questions and to sense-check data. Yet the Home Office is again moving at pace to adopt AI facial age estimation to assess asylum seekers’ ages, without any clarity about the way this tool has been trained on children who have suffered trauma. The vulnerable will bear the risks.

The last theme I want to briefly mention is the way that personal use of AI has the potential to turn off critical faculties and undermine relationships. Answers generated by AI tools carry the appearance of authority, often coupled with high levels of sycophancy. Do we really understand the impact of this sycophancy on decision-making? We do not understand it well. A recent scientific study published in Science sought to evaluate it and found that AI affirmed users’ actions 49% more often than humans, even when queries about personal dilemmas involved deception, illegality or other harms. In some experiments, even a single interaction with a sycophantic AI reduced participants’ willingness to take responsibility and repair interpersonal conflicts, creating perverse incentives for sycophancy to persist.

We need a values-led approach to how we engage with AI, across all areas of life, that puts human beings at the centre. I shall be interested to hear the Minister’s response.