Reflection #125 (14th June 2026 at Essex Church / Kensington Unitarians)
I feel some trepidation in setting out to speak on this subject! AI is a hot topic, in a developing area, which requires some degree of technical sophistication to grasp, and which people seem to have very strong and diverging opinions about! So I feel I’m stepping into a bit of a minefield here but I’ll give it my best shot… My one, flimsy, credential for speaking on such a topic is that I did take a module on AI and machine learning as part of my MSc in medical engineering down the road at Imperial – I even coded my own tiny little neural network! – but that was nearly 30 years ago. And the landscape has changed so much since then that it might as well have been 1000 years ago. (And I feel the need to say that though this was billed as a mini-reflection it’s turned into a maxi-reflection as it’s a big old subject with a lot of key points to mention)… But with all those caveats out of the way, let’s dive in.
On Friday night I went to a seminar on ‘The Spirituality of AI’ hosted by the London Jesuit Centre and one of the useful things I picked up there was a definition of artificial intelligence that was a bit broader than the one I went in with. One of the speakers, John-Clark Levin, stated that AI is any machine or a piece of software that performs functions that human intelligence would otherwise do. Historically, that includes things like calculators, where humans worked out the algorithm to do a particular task and then encoded it into a machine. But AI as we know it today is based on machine learning, and one key characteristic of this is that modern AI learns by itself, we don’t exactly instruct it; it finds patterns in data, in a way which is beyond our human ability to grasp, and functions like a ‘black box’. By this, I mean, we train AI on a bunch of inputs and it will give us an output that feels plausibly like the sort of thing human intelligence could have come up with, but we can’t reverse-engineer it to understand its reasoning, so we don’t know exactly how it got there.
A couple of years ago, in 2024, our Unitarian General Assembly passed a resolution on the subject of AI. It was put forward by Andi Phillips, minister with Upper Chapel in the centre of Sheffield, who’s been one of our strongest voices on this matter, and she has an academic background in engineering and maths so she knows what she’s talking about, she speaks the language. I’m going to share an abridged version of that resolution so you know what the official ‘party line’ is (in as much as Unitarians ever recognise such a thing). The resolution says: ‘This General Assembly of Unitarian and Free Christian Churches recognises the rapidly growing significance of automation, algorithms, and artificial intelligence, and encourages a balanced response mindful of both substantial benefits and threats. In particular it calls upon individuals and congregations to learn about AI, algorithms and their societal impact (whether good or bad), encourages Unitarian bodies to host and/or facilitate wider societal and philosophical conversations; and recognises that injustices, often affecting the already disadvantaged, are occurring through automation, and calls upon Unitarians to challenge such injustices.’
So, that was the first collective Unitarian pronouncement on AI, from a few years ago. Personally, I think the tone of that resolution is about right – it reflects my own mixed feelings with its talk of ‘benefits and threats’ and ‘impacts (good or bad)’ – the tone is not that of the Luddite or the dinosaur – it’s not a knee-jerk anti-AI reflex in opposition to the new – but it is a call to be appropriately cautious and critical.
Let’s start with some of the positives: I think it’s fairly self-evident that AI has tremendous potential to do good. There are undoubtedly many tasks that it can be useful for – and maybe in a way that could be positively transformative for humanity – this technology could be harnessed for the good. It can be used to synthesise complex information and make it more digestible; I’ve certainly heard people speak convincingly of AI as being a tool for accessibility.
Because of my background in medical imaging I immediately think of the ways in which it could be used to assist with diagnosis, drug development, treatment planning and so on. There are applications of AI in materials science which apparently open up possibilities for more efficient sources of green energy. It seems likely that there are many scientific and medical conundrums, where the obstacle to finding a solution is the sheer complexity and scale of the data, and in these cases AI could provide the support to enable significant breakthroughs. There are many ways in which AI could be used to support human flourishing.
But. Is that how we are collectively actually using AI right now? It seems – and I am trying not to get on my Victor Meldrew type soapbox about it – but it does seem that AI is being pushed onto us willy-nilly by tech companies and incorporated by default into all sorts of settings where it doesn’t seem necessary and may not be beneficial. It’s seamlessly integrated into Google searches, social media feeds, and any number of apps that many of us use all the time for work and play. And it’s becoming very apparent that AI isn’t a value-neutral technology – there’s so much scope for biases to be built in – and at the same time, the advent of Large Language Models like ChatGPT means that interacting with an AI these days feels much like talking to a human. This makes it too easy to be lulled into a false sense of security, to be charmed, to assume benevolence. Indeed, one of the recognised issues of AI is the problem of ‘sycophancy’. Even if there’s not someone of ill intent behind the AI (and that’s a big if), even if nobody is consciously trying to build in bias and use it for misinformation or propaganda, it seems that – by default – AI wants to tell us what we want to hear. And that in itself can be a dangerous thing. It can reinforce and radicalise people’s worldviews in an increasingly extreme way. And there’s even a known phenomenon of ‘AI delusion’ where people gradually come to trust what the (sycophantic) AI tells them in preference to the more complex, diverse, nuanced, or challenging voices they might hear in the world outside. There seem to be quite a number of cases where this has led people to lose touch with reality and behave in disturbing ways which cause great harm to themselves or others.
There are some less-serious ways in which this plays out too. If you use Google to search the internet, you’ll be aware that these days you are by default presented with an ‘AI overview’ before your old-fashioned search results, and if you look carefully, at the bottom of this answer to your query, there is a bit in very small print which says ‘AI responses might include mistakes’… But it can be a bit more than a mistake. This week I was listening to my favourite podcast, Three Bean Salad, and one of the hosts, Ben Partridge, described something that had happened to him on a recent holiday in Italy. As he travelled around, each day, he asked AI what events were going on locally. When he got to Rome, AI told him that there was a brass band competition going on in a public square, which is just exactly the sort of thing he would like, so he walked 40 minutes to get to the piazza where this brass band event was meant to be happening, only to find… there was no such competition. It was a complete fabrication. Fictitious. AI just made it up. Told him what he wanted to hear. And in such a confident and articulate way that he wouldn’t think to question it. Now that’s not a very consequential mistake. But what about when people ask AI for medical advice? Or use AI as a stand-in for a therapist? Then such ‘mistakes’ – where the AI just makes something up, confidently says what it thinks you want to hear, regardless of whether it has any basis in reality – they could be disastrous.
There are of course many ethical objections to AI too – I suspect most of us are pretty familiar with the sort of thing I’m talking about – the first ethical problem I became aware of was when the firms developing various AI models were instructing them to scrape (i.e. steal) various creative works – art, music, writing – and indeed our own personal data and even our likenesses – in order to use them to generate ‘new’ works. And then people use these ‘free’ AI-generated models to churn out images or words instead of paying creative workers, so they are exploited twice over. And this is a repeated pattern of getting humans to train AI models and then sacking them and replacing them with AI. Putting huge swathes of people out of work. And accelerating the concentration of wealth and power in the hands of a few.
Then there’s the environmental cost – the power and water needed for these enormous data centres – which we might think is worth it if the AI is being used for scientific or medical purposes that benefit humanity – but is it worth it in order for a billion people to generate an AI avatar of themselves as an action figure to share on Facebook (or whatever the latest meme doing the rounds might be)?
And what about the use of AI to create realistic photos and videos which are used maliciously to spread misinformation? We are accustomed to relying on such footage as ‘proof’ on which we base our opinions about what’s going on in the world. But it is getting increasingly difficult to tell what’s real and what’s made up. We’ve got a harder job on our hands to be discerning about what we see – can we believe our own eyes? – and that can drive us to be distrustful across the board. I suspect that such a climate, where people can’t reliably tell what is real, helps to drive people towards conspiracy theories and creates a distorted worldview.
All this may seem relentlessly negative. But it’s important to be clear-eyed about what we’re facing and the impact it is already having on human flourishing. The genie is out of the bottle now – there’s no going back – so how do we respond?
Unusually, both the resolution from the Unitarian GA, and the Pope’s encyclical, seem to be coming from a similar place. They say: we have to recognise that AI, like any human invention, can be used for good or ill. We can’t really escape it, or the effects of it, as it is being woven into all aspects of our lives. So let’s try and engage critically with it, harness it, shape it for the good, or at the very least learn what it is we’re dealing with, so that we can mitigate some of its harmful effects. It’s a question of discernment: when and how is it wise to make use of AI? I’ve been reminded of the concept of ‘epistemic hygiene’ – very useful in this day and age – the idea that we should take responsibility for protecting ourselves from misinformation and bias – being alert to the source of any information we’re presented with and not taking it at face value – and using our critical faculties (rather than being seduced by the sycophantic voice which mirrors our desires).
The spiritual imperative is for us to live our life, to pay attention, and reconnect with what’s real, rather than escape reality or fall into delusion. To use our unique gifts in the service of love, justice, and peace – as we so often say – rather than delegating the tasks of life to AI – it is for us to wrestle with these creative tasks, which help give us meaning and purpose, and grow and deepen our souls, over the course of a lifetime. We are called to express what’s authentically within us, as a basis on which to connect meaningfully with others. I came across a poem by Joseph Fasano, ‘For a Student who Used AI to Write a Paper’ which includes these lines: ‘I hear you. I know this life is hard now. I know your days are precious on this earth. But what are you trying to be free of? The living? The miraculous task of it?’
As our opening words today reminded us, ‘sometimes the only thing we can do is be still for a moment to remind ourselves what is real’. Perhaps that’s the thing we should hold onto, and keep bringing ourselves back to, in this ‘age of artifice’. Like James Crews said, let’s write our own sentences, make cake, walk in the woods, have wild encounters. Or as Jo Atkins-Potts hinted, in our poem for meditation, let’s embrace the slow and messy human processes of love and loss, and be fully present to each other. May it be so, for the greater good of all. Amen.
Reflection by Jane Blackall