Article
AI - Artificial Intelligence
Creed
Morality
Spiritual formation
6 min read

The moral machine: algorithms that give a window into the soul

In TikTok’s algorithm Graham Tomlin saw something that got him thinking. Could it lead to moral health rather than harm?

Graham is the Director of the Centre for Cultural Witness and a former Bishop of Kensington.

A abstract grid of colourful cubes with arrows, crosses and cubes, viewed from above and at an angle
Champ Panupong Techawongthawon's illustration of artificial intelligence.
Google DeepMind on Unsplash.

A few years ago, I was thinking of buying a camera for my wife as a birthday present. I lazily browsed a couple of websites to check out the options. Then something odd started to happen. Somehow, my laptop seemed to think this was a good idea and sprang into action. Whenever I went onto Amazon, Ebay or any other website selling stuff, it kept pushing adverts for cameras at me. Canon or Nikon? Point and Shoot or DSLR? How did it know? Could it read my mind?

It was the first time I noticed the power of the algorithm.

A bit later on, thinking I ought to get up to speed with the regions of social media that I had little clue about, I opened up TikTok and started to swipe upwards (apparently that seemed to be the way to do it). This time I was determined not to like or unlike anything, follow anyone or be followed by anyone. Yet mysteriously, it still worked out what I liked and kept pushing short, addictive videos at me, enticing clips of football, mountains and music along with other random stuff mixed in. How did it know me so well?

It is as old as the hills. The algorithm simply takes the desires of your heart and amplifies them.

Of course, better informed people than me know how all this works. The algorithm figures out which accounts you follow, any comments you’ve posted, clips you’ve liked or shared, and in particular, videos you watched all the way to the end. So, if you linger over a video, it knows you like it. If you rush on quickly to the next one, it makes a mental note that you’re not so keen.

It all feels a little sinister, yet very clever. You often read dark theories of social media, and the way it is re-wiring our brains. Yet when you look a little closer, it is as old as the hills. The algorithm simply takes the desires of your heart and amplifies them. So, if you like or linger over certain videos expressing a particular cultural or political opinion it will send you more of the same. The result is we get confirmed in our own frameworks which never get challenged by others. It's part of why we are so polarised as societies these days. When you ask why ‘the other side’ cannot see the obvious truth that you see, the answer is that they literally don’t see it. They don't see it because the algorithm doesn't feed them the same things as it feeds you.

As a result, TikTok or Facebook is an alarming mirror into the soul – see what it sends you and it just may be that it tells you more about yourself than you would like to know.  

Social media like TikTok, Facebook and Twitter (or X)  learn to recognise what your heart really desires (not just what you say you do). They notice what you linger over, what catches your fancy and sends you more of the same. They are, apparently, studiously neutral on moral questions. They seem to have no moral designs on you to school or form your soul in particular ways, but are simply a reflection of your own longings. What TikTok, Facebook, Instagram and the others all do is to propel you further in the moral direction in which you are already headed. Which for most of us, is not a great idea.

Now of course there would be howls of protest if TikTok announced a moral code – that it was about to encourage virtue and discourage vice by deliberately sending us improving videos, material that the mysterious people who run it think is good for us. And that is not because we think virtue is bad and vice is good, but because we can’t decide on what virtues we want to encourage or what vices to stamp out. We draw a line at cruelty to children and extreme violence, but not much else. It is also because we hold as sacrosanct the freedom of the (adult) individual to choose his or her own way in life, as long as they don’t hurt anyone else.

Such sites are examplars and vehicles of expressive individualism – not just in the myriads of people who show off their dance moves, sing their songs or act out half-funny scenes on a golf course, but in that they confirm me in me my own wishes. They don’t tell me what to want. But they give me more of what I want. As a result, TikTok or Facebook is an alarming mirror into the soul – see what it sends you and it just may be that it tells you more about yourself than you would like to know.

It matters what we feed our souls with. It matters what stories we allow ourselves to be told.  

Such sites appear to be morally neutral. They don’t seem to aim to educate or form you in any particular direction. Or at least they are supposed not to. But of course nothing is entirely neutral.

Funnily enough, it’s not how we bring up children, or educate ourselves. When we bring up a child, most of us have some kind of vague or not so vague moral code in mind. We reward kind and helpful behaviour, and we punish selfish and mean actions. We don’t tend to give more of the same to a child who has eaten the first half of the packet of biscuits, or encourage a brother to hit his sister yet again. We have a goal of some form of moral formation in mind.

Yet, despite our confusion over which virtues to encourage, we need some kind of moral guidance for our wandering and flawed hearts, linked to eyes that are tempted to feast on things that fascinate but are not good for us. Like a glutton who cannot stop eating, even if these sites don’t themselves push extreme violence, pornography, aggression, they offer enough of the soft version of these to draw you in. And it’s not hard to find sites that will take you deeper into the darkness. And those sites will already know the way you are thinking and desiring and are ready to pull you in deeper into the mire.

The problem is not so much with the algorithm. It is with us. Netflix’s documentary, ‘The Social Dilemma’ quotes an alarming statistic - that fake news spreads six times faster than the truth. The reason is not hard to find. We are fascinated by the sensational and alarming rather than something a little more ordinary yet which happens to be true. As one person in the documentary put it: “The internet has a bias towards false information. Because false information makes more money. The truth is boring.”

The moral philosopher Gilbert Meilaender wrote:

“Successful moral education requires a community which does not hesitate to inculcate virtue in the young, which does not settle for the discordant opinions of alternative visions of the good, which worries about what the stories of its poets teach.”

It matters what we feed our souls with. It matters what stories we allow ourselves to be told.

The purveyors of social media are not innocent in this as they do exploit our worst tendencies, but in the end they simply confirm us in our own moral confusion. Yet it does point up the problem in the liberal ideal of leaving ethical decisions entirely up to the individual, to give entirely free choice without any guidance, because with our crooked hearts, it will always end up feeding the darker sides of our characters without a corresponding pull in the other direction, something which Christians called divine Grace.

St Paul wrote to the small group of Christians in Philippi, surrounded by the highly sexualised and violent culture of the Roman empire: “whatever is true, whatever is honourable, whatever is just, whatever is pure, whatever is pleasing, whatever is commendable, if there is any excellence and if there is anything worthy of praise, think about these things.”

I’m not saying don’t watch TikTok. But here’s an idea. Why not try to make it into a morally forming version of what you want to be, not what you are? Exercise a bit of moral direction yourself. If you see a video which you know in your conscience is not good, or is spreading lies, swipe it away quickly. If you see something positive, dwell on it.

If you approach it this way, you might just be able to persuade the algorithm to shape you in good ways and not the bad. It could become a means of growing in goodness, but only if you want it to be.

Article
AI - Artificial Intelligence
Creed
Wisdom
6 min read

Forget AI: I want a computer that says ‘no’

Chatbots only tell us what we want to hear. If we genuinely want to grow, we need to be OK with offence

Paul is a pioneer minister, writer and researcher based in Poole, Dorset.

A person hold their phone on their desk, a think bubble from it says 'no'.
Nick Jones/Midjourney.ai.

It is three years since the public release of Open AI’s ChatGPT. In those early months, this new technology felt apocalyptic. There was excitement, yes – but also genuine concern that ChatGPT, and other AI bots like it, had been released on an unsuspecting public with little assessment or reflection on the unintended consequences they might have the potential to make. In March 2023, 1,300 experts signed an open letter calling for a six month pause in AI labs training of the most advanced systems arguing that they represent an ‘existential risk’ to humanity. In the same month Time magazine published an article by a leading AI researcher which went further, saying that the risks presented by AI had been underplayed. The article visualised a civilisation in which AI had liberated itself from computers to dominate ‘a world of creatures, that are, from its perspective, very stupid and very slow.’ 

But then we all started running our essays through it, creating emails, and generating the kind of boring documentation demanded by the modern world. AI is now part of life. We can no more avoid it than we can avoid the internet. The genie is well and truly out of the bottle.  

I will confess at this point to having distinctly Luddite tendencies when it comes to technology. I read Wendell Berry’s famous essay ‘Why I will not buy a computer’ and hungered after the agrarian, writerly world he appeared to inhabit; all kitchen tables, musty bookshelves, sharpened pencils and blank pieces of paper. Certainly, Berry is on to something. Technology promises much, delivers some, but leaves a large bill on the doormat. Something is lost, which for Berry included the kind of attention that writing by hand provides for deep, reflective work.  

This is the paradox of technology – it gives and takes away. What is required of us as a society is to take the time to discern the balance of this equation. On the other side of the equation from those heralding the analytical speed and power of AI are those deeply concerned for ways in which our humanity is threatened by its ubiquity. 

In Thailand, where clairvoyancy is big business, fortune tellers are reportedly seeing their market disrupted by AI as a growing number of people turn to chat bots to give them insights into their future instead.  

A friend of mine uses an AI chatbot to discuss his feelings and dilemmas. The way he described his relationship with AI was not unlike that of a spiritual director or mentor.  

There are also examples of deeply concerning incidents where chat bots have reportedly encouraged and affirmed a person’s decision to take their own life. Adam took his own life in April this year. His parents have since filed a lawsuit against OpenAI after discovering that ChatGPT had discouraged Adam from seeking help from them and had even offered to help him write a suicide note. Such stories raise the critical question of whether it is life-giving and humane for people to develop relationships of dependence and significance with a machine. AI chat bots are highly powerful tools masquerading behind the visage of human personality. They are, one could argue, sophisticated clairvoyants mining the vast landscape of the internet, data laid down in the past, and presenting what they extract as information and advice. Such an intelligence is undoubtedly game changing for diagnosing diseases, when the pace of medical research advances faster than any GP can cope with. But is it the kind of intelligence we need for the deeper work of our intimate selves, the soul-work of life? 

Of course, AI assistants are more than just a highly advanced search engines. They get better at predicting what we want to know. Chatbots essentially learn to please their users. They become our sycophantic friends, giving us insights from their vast store of available knowledge, but only ever along the grain of our desires and needs. Is it any wonder people form such positive relationships with them? They are forever telling us what we want to hear.  

Or at least what we think we want to hear. Because any truly loving relationship should have the capacity and freedom to include saying things which the other does not want to hear. Relationships of true worth are ones which take the risk of surprising the other with offence in order to move toward deeper life. This is where user’s experience suggests AI is not proficient. Indeed, it is an area I suggest chatbots are not capable of being proficient in. To appreciate this, we need to explore a little of the philosophy of knowledge generation.  

Most of us probably recognise the concepts of deduction and induction as modes of thought. Deduction is the application of a predetermined rule (‘A always means B…’) to a given experience which then confidently predicts an outcome (‘therefore C’). Induction is the inference of a rule from series of varying (but similar) experiences (‘look at all these slightly different C’s – it must mean that A always means B’). However, the nineteenth century philosopher CS Pierce described a third mode of thought that he called abduction.  

Abduction works by offering a provisional explanatory context to a surprising experience or piece of information. It postulates, often very creatively and imaginatively, a hypothesis, or way of seeing things, that offers to make sense of new experience. The distinctives of abduction include intuition, imagination, even spiritual insight in the working towards a deeper understanding of things. Abductive reasoning for example includes the kind of ‘eureka!’ moment of explanation which points to a deeper intelligence, a deeper connectivity in all things that feels out of reach to the human mind but which we grasp at with imaginative and often metaphorical leaps.  

The distinctive thing about abductive reasoning, as far as AI chatbots are concerned, lies in the fact that it works by introducing an idea that isn’t contained within the existing data and which offers an explanation that the data would not otherwise have. The ‘wisdom’ of chatbots on the other hand is really only a very sophisticated synthesis of existing data, shaped by a desire to offer knowledge that pleases its end user. It lacks the imaginative insight, the intuitive perspective that might confront, challenge, but ultimately be for our benefit. 

If we want to grow in the understanding of ourselves, if we genuinely want to do soul-work, we need to be open to the surprise of offence; the disruption of challenge; the insight from elsewhere; the pain of having to reimagine our perspective. The Christian tradition sometimes calls this wisdom prophecy. It might also be a way of understanding what St Paul meant by the ‘sword of the Spirit’. It is that voice, that insight of deep wisdom, which doesn’t sooth but often smarts, but which we come to appreciate in time as a word of life. Such wisdom may be conveyed by a human person, a prophet. And the Old Testament’s stories suggests that its delivery is not without costs to the prophet, and never without relationship. A prophet speaks as one alongside in community, sharing something of the same pain, the same confusion. Ultimately such wisdom is understood to be drawn from divine wisdom, God speaking in the midst of humanity   

You don’t get that from a chatbot, you get that from person-to-person relationships. I do have the computer (sorry Wendell!), but I will do my soul-work with fellow humans. And I will not be using an AI assistant. 

Support Seen & Unseen

Since Spring 2023, our readers have enjoyed over 1,500 articles. All for free. 
This is made possible through the generosity of our amazing community of supporters.

If you enjoy Seen & Unseen, would you consider making a gift towards our work?

Do so by joining Behind The Seen. Alongside other benefits, you’ll receive an extra fortnightly email from me sharing my reading and reflections on the ideas that are shaping our times.

Graham Tomlin
Editor-in-Chief