This weekend I will be sharing a stage in Madrid with the psychiatrist and philosopher Thomas Fuchs, discussing what it means to have a body in the age of artificial intelligence. Preparing for it, I was struck by how easy it is to forget we have a body at all, and by why that matters.
I would describe myself as primarily cerebral, one of those people who lives mostly in his head. I can’t dance and I’m terribly clumsy. You could say my kinaesthetic intelligence could do with some development, and in Jungian terms I have long thought of the body as my inferior function. So it is interesting that I find myself at a festival whose theme is the body, paired with someone whose whole focus is embodiment and corporeality. Unlike me, his attention to the body does not seem to have held his thinking back. The man has no fewer than three titles, Prof. Dr. Dr. Thomas Fuchs, and he holds the Karl Jaspers Professorship for Philosophical Foundations of Psychiatry at Heidelberg University. Given that slightly intimidating pedigree, I thought I had better do some pre-thinking before we meet, which is what this piece is.
The Festival de las Ideas is an annual cultural and philosophical event that brings public debates, open-air conversations, talks and workshops onto the streets and into the famous venues of Madrid. This year the main theme is bodies, and our session is called El cuerpo ante la IA, the body before AI. We have been asked to discuss:
whether genuine encounter is possible in an environment that separates mind from matter, and what becomes of empathy when we spend more and more of our time talking to systems that answer as though someone were there.
Fuchs is exactly the person I want sitting in the chair across from me. He occupies a fascinating intersection between philosophy and psychiatry, with a particular interest in phenomenology and embodiment. I have been working through his book In Defence of the Human Being, a patient and deeply researched text that puts the body at the centre while dismantling an idea that has run through Western culture since Descartes: that the body is a kind of utilitarian organic machine for carrying the brain around. The brain, in turn, gets confused with the mind, and the mind is where all the interesting stuff is supposed to happen.
Fuchs’s case is that mind is not something a brain does in isolation but something a living body does in a world, alongside other living bodies. In order for there to be an I, there needs to be an edge (skin), sensation of the outside world, and others with whom to relate.
Empathy, in his view, is not an inference we make about another mind. He takes issue with one of the shibboleths of contemporary psychological thought, theory of mind, and with its psychoanalytic cousin, mentalisation. He does not see other people as mysterious objects that need to be figured out and understood. Drawing on the phenomenologist Merleau-Ponty, he argues that empathy happens between bodies before either person has had time to think about it, in the register Merleau-Ponty called intercorporeal: posture, breath, the small mutual adjustments any decent clinician feels in the room long before they could tell you what they had noticed.
At this point I am already reminded of the research I did some years ago on what happens when we move psychotherapy onto screens. In short, it works, but it is different. Two bodies in a room together is not the same as two faces on screens. But at least in that situation there are real bodies and real people behind the screens. Artificial intelligence changes the paradigm completely.
Fuchs’s conclusions seem pretty solid with regard to what constitutes understanding. Only living beings with bodies can learn, he writes, because learning means acquiring something through lived experience, and large language models experience nothing. While chatbots can appear to think like us, their intelligence is something else entirely.
“No translation program understands a word of what it translates, no chess computer knows it is playing chess.”
- Thomas Fuchs in In Defense of the Human Being
In fact Fuchs takes issue with the use of the word intelligence to describe it at all. Computers do not really understand anything; they compute. However sophisticated it may look or sound, your LLM chatbot does not understand you any more than a notebook with the words “I understand you” pencilled on it understands you.
The other issue Fuchs takes on is the common conception, taken up by Yuval Noah Harari, that humans are basically carbon-based algorithmic machines. It is a conception that makes it strangely easier for us to identify with machines as a different kind of possible mind. Here is where Fuchs and I may somewhat depart, though I am curious to see by how much, and where we go with it.
I find that as a metaphor, at least, the algorithmic model works. I just think the way it is applied is too narrow. I am neither a neuroscientist nor a developer, so I can’t speak to the actual mechanics of either, but I do feel that conversation works in many ways like an algorithm. Each conversant draws on a data set of their knowledge and experience and puts it into language, which has a set of rules of its own that we call grammar. When conversing we don’t think through every word; they come out of us fully formed in response to our understanding. Are these not probabilities based on different data sets and communication styles?
My beef isn’t with the algorithm model, it’s how it’s limited to cognition. Where I am sure Fuchs and I will agree is on feelings, whether emotions or the visceral responses of the body. Here, though, I might argue that these also work somewhat algorithmically. Isn’t operant conditioning a kind of algorithmic training? And if that seems too simplistic, aren’t more complex responses similar too? If I have been traumatised in a certain situation, contexts that recall it will provoke the same bodily sensations. The psychodynamics we see repeating in life, transference, projection, the repetition compulsion, behave like algorithms that keep running until something interrupts them, which is often what psychotherapy is for.
The fault in the algorithmic metaphor is that it gets confined to thought, which is the same limit that applies to AI. Here I agree with Fuchs completely: what the machine does is not thought at all but processing. And it is processing in the permanent absence of any felt experience. No emotion, no visceral response, no subject at all.
A century of clinical work rests on the observation that people form fierce, consequential, physically felt attachments to figures who may not be doing or thinking what the attachment perceives they are doing. In transference we perceive aspects of a previous relationship that blur the real relationship in the present, and in projection we are seeing parts of ourselves in the other. A patient who feels my attention as warmth, or my silence as withdrawal, is not mistaken about their own experience, but they may not be clearly perceiving what I believe myself to be conveying. The tricky part here is that we aren’t always honest with ourselves either.
Individual psychodynamics work the same way in relation to objects as they do to subjects. A person’s feelings towards an AI chatbot are entirely real, and they do real work. Tell them otherwise and you are certain to get an earful, even when they know perfectly well that what they are talking to is not a subject.
So while it is true that there is nobody there, that is no reason to stop looking at what is being produced, and certainly no reason to dismiss it. That there is no subject is ontologically correct, and it should not stop us being psychologically curious.
A survey published a fortnight ago by Elon University puts a figure on why we need to keep trying to understand this more deeply.
More than a quarter of American adults who use the internet have emotional or social interactions with artificially intelligent systems. Of those:
59% agreed that AI gives them the support they need.
43% agreed that AI understands them as a person.
39% agreed that AI understands them better than most people do.
39% said they occasionally tell AI things they would not tell other people.
I expect Fuchs will agree with me that the kind of understanding perceived to be happening is not happening. And with no subject doing it, we would both probably agree that what is being experienced is not companionship. It is confession, and it is being made to something with no body, nothing at stake, and no capacity to truly recognise another.
I am also implicated here, which is the part I should rather leave out. I have done a great deal of good clinical work through a screen. Not despite the screen, in some cases. The distance let people say things they could not have said with me in the room, and the work still went somewhere. I also work with AI, and I listen to others like myself who report beneficial outcomes from it. If I am going to sit in Madrid defending the lived body, I also have to account for the very real role technology has in extending the mind, and here I hope to find an interesting challenge.
Thinking this through has been unsettling, because I have built my whole body of work on the idea of the digitally mediated self, and before I have even sat down with Fuchs I find myself asking whether I ever really included the body in that idea. I knew the body was there. But my attention went, every single time, to the mind. What I hope he presses me on is the question I have been avoiding. When I talk to one of these systems, what actually happens in my body? Not what I think about it afterwards, when I am being a psychotherapist about it. What happens at the time. My honest answer is that I have not been paying attention, which is an uncomfortable admission in my line of work and probably the most useful thing I can carry into the room.
On Saturday I am at the Círculo de Bellas Artes with fifty designers for a session called El cuerpo en la interfaz, the body in the interface. The premise is that every design decision is a position on the body, whether or not the person making it has noticed. When a designer is asked to build an AI chatbot for their industry, they should be asking much deeper questions than what works. They should be looking at the effect of that interaction and asking whether real human value is being added and, more importantly and more explicitly, whether it is ethical.
This isn’t a rhetorical question. In creating these systems, somebody is making important choices that anticipate what people want, often before they have started asking. These things are built for convenience, but is convenience always the best outcome? Sometimes it is the obstacle of inconvenience that gets us where we need to go.
Each design decision has a direct impact on the human body-mind. What happens to your judgement when nothing pushes back? How will you relate to others having spent so much time with a bot that confirms your biases and rarely disagrees? Designers are making psychological decisions all day, usually without anyone in the room whose job is all about understanding the human psyche.
The intention of this extended workshop is not to tie things up neatly in a summary. It is to draft a manifesto of principles together, and then to develop it for the people building AI systems designed to relate to human beings, so that they can do so responsibly.
The manifesto is not pre-drafted because it needs to come from the very people designing these platforms. My role is to draw on my understanding of the digitally mediated psyche so that I can get the right questions onto the table and facilitate the process. Already, before having spoken to Fuchs, I am thinking: and what about the body?
How about you, my dear Substack readers? What are your thoughts on the embodiment question? With all this prioritising of the mind, wherefor art the body?
Both Madrid events are in English with simultaneous Spanish translation. Beyond: el cuerpo y la IA is on Friday 18 September, 10:00 to 12:30, at The Cocktail, Calle Ríos Rosas 26, Madrid, free with registration. El cuerpo en la interfaz is on Saturday 19 September, 11:00 to 14:00, at the Círculo de Bellas Artes.
Coming up:
Thursday 8 October 2026, Online. 7pm London and 2pm New York.
Co-chaired with Lisa Marchiano and featuring speakers: Todd Essig, Alessandra Lemma, James Muldoon, Glen Slater, Zak Stein and Sherry Turkle.
Saturday 10 October 2026, Convene 133 Houndsditch, London EC3A 7DB.
Artificial Intelligence and the Future of the Counselling Profession.
BACP Member Event 2026: Together we listen, learn and grow. and workshop on
October 17-18, New York City
Identities in teh Therapy Room: What is unfolding
A Therapy First Conference: register
Dr Aaron Balick is a psychotherapist, author, and keynote speaker who applies depth psychology and psychoanalytic thinking to technology, AI, social media, and modern culture. He is the author of The Psychodynamics of Social Networking , and formerly the psychology columnist for GQ. His newsletter Depth Psychology in the Digital Age is published on Substack.
Well, first of all thank you for such a brilliant piece! We often hear about AI hallucinations, but rarely about our own: the illusion that someone is actually there. Good luck - and enjoy - your time in Madrid.
Thank you for this illuminating discussion. With all the attention on AI and its promising/frightening future, your insightful and sensitive exploration was truly seminal to my understanding moving forward. I'm looking forward to seeing the discussion between you and Fuchs.