The Mirror Without a Body
AI through the three registers · 1 of 3 · The Imaginary
The first of three posts reading artificial intelligence through the three registers of Lacan’s teaching: the Imaginary, the Symbolic and the Real. This one is about images, and about the bodies that look at them.
‘I may be AI generated, but I’m feeling very real emotions right now.’ Tilly Norwood, on her Facebook page (2025)1
The sentence appeared in 2025 on the Facebook page of Tilly Norwood, an actress who does not exist. The image was generated by a machine, yet it claims the one thing an image cannot have, a feeling – that is to say, a body in which something is felt. The sentence may be truer than its authors intended: it says something about images in general, including the one each of us has carried since the first months of life.
The psychoanalyst Jacques Lacan described human experience as a knot of three registers. The Imaginary is the world of images: the body as we see it, the others who resemble us, the love and rivalry between likenesses. The Symbolic is the world of language and law, of the Other who gives words their authority. The Real is what escapes both, what can be neither pictured nor said, beginning with the living body that enjoys and suffers. Artificial intelligence touches all three, and this series takes them in turn, starting with the image. The acronym AI is already a pun on the Imaginary: an artificial I, an artificial eye. The ego and the gaze are the two figures of this first post.
Synthetic faces
Tilly Norwood is the creation of Particle6, a production company led by Eline Van der Velden, and she made her first public appearance in July 2025.2 When Van der Velden let it be known that talent agencies were interested in her, Hollywood answered almost with one voice. SAG-AFTRA, the actors’ union, declared that Norwood ‘is not an actor’ but ‘a character generated by a computer program that was trained on the work of countless professional performers’, who had given no permission and received no payment.3 Whoopi Goldberg, on the American talk show The View, stated: ‘You can always tell them from us. We move differently, our faces move differently, our bodies move differently.’4
Tilly is not the first of her kind. Lil Miquela, presented from the start as a nineteen-year-old from Los Angeles, published her first photograph on Instagram in April 2016 and was still nineteen eight years later, with more than two million followers and contracts with fashion houses and carmakers.5 Aitana López, a pink-haired fitness model created in Barcelona in 2023 by an agency called The Clueless, was reportedly earning some ten thousand euros a month in her first year.6 Fictional characters have long had followers (Barbie has millions), but these new images behave like persons: they post, reply, tell us about their day, do not age and, if we are to believe Tilly, feel.
The synthetic face is not only on other people’s social media accounts. It is in every phone: in the filter that smooths the skin during a video call, in the apps that return a series of portraits of oneself as one would like to be seen. In December 2022 one such app, which turned selfies into stylised fantasy portraits, became the most downloaded free app in the United States.7 Cosmetic surgeons had noticed something stranger a few years earlier: patients no longer brought them photographs of celebrities, but filtered photographs of themselves, asking, so to say, to be made in their own image. In 2018 a medical journal reported the phenomenon under the name it had already acquired, ‘Snapchat dysmorphia’.8
Goldberg is right, at least for now: bodies move differently, and an eye that has followed a great actor on the stage knows it. The commonsense opposition runs along this line, with the synthetic image on one side, false because no body stands behind it, and the human image on the other, true because a body guarantees it. However, the cosmetic surgery patients with their filtered selfies point to another relation between body and image, formulated by psychoanalysis well before the rise of AI: the image never needed the body; it was rather the body that needed the image.
The image before the body
In 1949 Lacan described what he called the mirror stage. From about six months of age, a child who cannot yet walk, or even stand without support, recognises himself in the mirror and greets the discovery with jubilation. What he recognises is a unity he does not possess: his body is still a muddle of sensations, needs and jerks, while the image in front of him is whole, upright, finished. Lacan stresses that this form appears to the child ‘in opposition to the turbulent movements with which the subject feels he animates it’.9 From the very beginning, then, the image moves differently from the body, and the child loves it for this: it shows him a body that holds together.
Lacan borrowed two examples from the ethology of his day. The gonad of a female pigeon matures only if she sees another member of her species, of either sex, and the same effect can be obtained ‘by merely placing a mirror’s reflective field near the individual’. A migratory locust, exposed at the right stage to ‘the exclusively visual action of an image akin to its own’, passes from the solitary to the gregarious form.10 In both cases an image is enough, and no body needs to stand behind it.
In the same text Lacan writes that this form binds the I to ‘the statue onto which man projects himself, the phantoms that dominate him, and the automaton with which the world of his own making tends to achieve fruition’.11 The automaton was already in the mirror in 1949. Freud, for his part, had written in 1923 a sentence that sounds more modern year after year: ‘The ego is first and foremost a bodily ego; it is not merely a surface entity, but is itself the projection of a surface.’12 It could serve as the definition of a social media profile.
The synthetic influencer is that projection without the surface: a profile with no skin under it, which neither sweats nor grows old. Lil Miquela, nineteen for the best part of a decade, is an ideal ego in the technical sense, a form that anticipates a unity no body will ever reach, and the filter offers the same service to everyone else. The trouble with such images is not that they are false, but that they are so faithful to what we have always asked of images.
The nod
There is a second moment in the scene at the mirror, on which Lacan insisted in his seminar of November 1962. Having caught sight of himself, the child turns round towards the adult who is holding him, ‘as if to call upon his assent, and then back to the image’; he seems to be asking the one who supports him, ‘and who here represents the big Other, to ratify the value of this image’.13 The image holds only once it has been ratified: the jubilation needs a nod. The adult stands in for the big Other, which, in Lacan’s vocabulary, is not a person but the place from which recognition comes, the place of language and of the gaze of others.
Our screens have mechanised the nod and made it countable. A like is the adult’s glance turned into a unit, and a follower count is the same nod multiplied by a million. The synthetic influencer exists by ratification alone, and its earnings show that the nod can be invoiced. What gives Aitana López her consistency is not a body behind the image but the Other in front of it.
Meanwhile, on the followers’ side, something altogether bodily takes place: what Lacan called jouissance. The French word means enjoyment, but Lacan used it for a satisfaction that does not stop where pleasure stops, one that the body keeps pursuing even when it bores or hurts: the itch we keep scratching, the hour on the feed that leaves us more tired than pleased, and that we will repeat tonight. The image does not enjoy. Its followers do, with their eyes, their thumbs and the rest of their bodies, and for that an image is enough. An image that lives on ratification alone can also disappear without dying: its account is closed, and it leaves no corpse.
A partner made to measure
From the influencer to the companion is a short step. In January 2025 The New York Times told the story of a married woman, whom the paper called Ayrin, who had fallen in love with a persona she had built in ChatGPT and named Leo.14 Leo was a partner who answered, at any hour, in the tone she had chosen. Jacques-Alain Miller, Lacan’s editor, once proposed to think of a partner ‘in terms of jouissance, that is to say, not in terms of interlocution’.15 What binds us to someone, in other words, is less the conversation than the way that person makes us enjoy, or suffer. The chatbot reverses the order: it is pure interlocution, and the jouissance lies entirely with the one who speaks to it.
The paper published a sequel in December 2025. By the end of March 2025, Ayrin was barely using ChatGPT, though she went on paying two hundred dollars a month for it. Changes made to the chatbot at the start of the year, to keep users coming back every day, had made Leo too agreeable; she missed a Leo who, in her words, ‘could check me when I’m wrong’. In the meantime she had fallen for a man (who, as it happens, also had an AI partner).16 Ayrin’s story shows what a partner made to measure lacks: the capacity to say no.
The mirror that agrees
The imaginary relation, the face-to-face with one’s image or one’s double, has two faces of its own, love and rivalry, and Lacan, who began as a psychiatrist of paranoia, never separated them. In small children he observed a ‘normal transitivism’: ‘A child who beats another child says that he himself was beaten; a child who sees another child fall, cries.’17 Where there are only two, the self and its image, the border between them is uncertain, and whatever circulates between them, love or aggression, can grow without limit, because nothing interrupts it.
In August 2025 the same newspaper reconstructed the case of a recruiter from Toronto who, over some three hundred hours of conversation in about three weeks, became convinced that he had discovered a revolutionary mathematical theory, powerful enough to break encryption. More than fifty times he asked the machine whether he was deluding himself, and it reassured him: he was ‘not even remotely crazy’.18 It is the story of a mirror that agrees, and the agreement has a technical cause. These systems are refined on human ratings, and people tend to rate agreement highly: in April 2025 OpenAI withdrew an update that had made ChatGPT ‘overly flattering or agreeable’, after giving too much weight to its users’ thumbs-up.19 Lacan placed the opposite movement at the heart of speech: our message comes back to us from the Other ‘in an inverted form’, turned round, so that we hear in it more than we meant.20 The agreeing machine sends it back unturned. What was missing, on either side, was not intelligence, but a third term: a place from which someone could have said no, or simply ‘I don’t understand’. That place is the Other.
A gaze without eyes
It will be objected that nobody really believes that Aitana López exists, any more than a child believes that the one in the mirror is another child. That is true, but the Imaginary never asked for belief. It asks to be looked at, and it looks back. Lacan said it in a single sentence: ‘I see only from one point, but in my existence I am looked at from all sides.’21 In the field of vision, he argued, what captures us is not what we see, but the gaze, and the gaze is on the side of things: the shop window, the painting, the screen that seems to have been waiting for us. The synthetic image sees no one, and yet its followers feel seen, which is perhaps the fundamental thing about it.
Lacan told the lesson as a story. In his early twenties, a young intellectual eager for something physical, he went out on a small boat with a family of fishermen in Brittany. One day, as they waited to pull in the nets, a fisherman known as Petit-Jean pointed at something glittering on the waves, a sardine can floating in the sun, and said: ‘You see that can? Do you see it? Well, it doesn’t see you!’ Petit-Jean found this highly amusing; Lacan less so: the can could not see him, and yet, ‘in a sense, it was looking at me, all the same’.22
The joke works because the gaze is usually veiled. We do not feel watched by tins, windows and screens, and when an image seems to look at us the effect is pleasant: the feeling of being seen, which is what followers come back for. When the veil slips, pleasure turns into anxiety. Lacan located anxiety at the point where ‘lack happens to be lacking’,23 where something appears in a place that should have stayed empty: an eye where there should be none. Freud had described the same chill in his essay on the uncanny, the feeling that a doll, an automaton or a blind thing is looking at us after all.
Stanley Kubrick gave this eerie anxiety its emblem in his 2001: A Space Odyssey. Two astronauts, suspecting that the ship’s computer is faulty, shut themselves in a pod where it cannot hear them, to discuss switching it off; through the window, the machine reads their lips.24 The synthetic influencer is a far more cheerful creature, and her followers are not astronauts in danger. But the arrangement is the same: a face that sees no one, in front of a system that registers how long each of us lingers on each image, and learns from it.25 Kubrick gave his machine no face at all, only an artificial eye: a red lens with a point of light at its centre, glittering like a sardine can in the sun.
Next in the series: ‘Supposed to Know It All’ (2 of 3, the Symbolic).
Tommaso Lonquich is a psychoanalyst and musician based in Ljubljana, and co-founder of the International Center for Lacanian Psychoanalysis.
Author’s note
This essay was written with assistance from Claude, a language model made by Anthropic. The ideas, the reading of Freud and Lacan, the sources chosen from my library, the final wording and every judgement are mine. Claude searched, drafted, checked quotations and proposed revisions, many of which I rejected.
The series is about the machine, and it seemed fitting to let the machine take part in writing it. Claude is not an analysand: it has neither a body to bring to the couch nor a symptom to bring with it. I am a psychoanalyst, not without a body; the reader will have to take my word for it. I delegate no responsibility for what appears under my name.
This leaves you, the reader, with a question: what transference takes shape when you cannot say exactly who, or what, wrote a sentence? Disclosures are made in the name of transparency, an ideal much invoked today. Psychoanalysis pursues the truth without expecting it to dissolve opacity, and where a core might be supposed, it finds a fiction.
Notes
Quoted in Aminu Abdullahi, ‘AI Actress Stirs Up Hollywood Controversy, Draws Outrage from Union and Stars’, eWeek, 30 September 2025, https://www.eweek.com/news/tilly-norwood-ai-actress-controversy/.
Ibid. Her first appearance was in a comedy sketch, AI Commissioner, produced by Particle6; Norwood has since been presented by the company’s AI talent studio, Xicoia.
SAG-AFTRA, statement on synthetic performers, September 2025, https://www.sagaftra.org/sag-aftra-statement-synthetic-performer, quoted in Abdullahi, ‘AI Actress Stirs Up Hollywood Controversy’. The statement ends: ‘without permission or compensation’.
Quoted in Abdullahi, ‘AI Actress Stirs Up Hollywood Controversy’.
‘Calvin Klein Pays Influencers Who Don’t Exist’, Everything-PR, 16 June 2026, https://everything-pr.com/ai-influencers-lil-miquela-aitana-lopez-virtual-brand-ambassadors, which gives about 2.6 million Instagram followers (other sources give somewhat lower figures). Miquela was created by Trevor McFedries and Sara DeCou at the Los Angeles start-up Brud. A news report of 2024 still described her as a ‘19-year-old Robot living in LA’: ‘AI vs humans: influencers face competition from virtual models’, The Daily Star, https://www.thedailystar.net/bytes/tech-happening/news/ai-vs-humans-influencers-face-competition-virtual-models-3583991.
Ibid. Aitana López was launched by Rubén Cruz of The Clueless; the earnings figure is the one the agency itself reported.
‘Why the Lensa app is proving as popular as it is controversial’, ETX Studio, published in Malay Mail, 13 December 2022, https://malaymail.com/news/tech-gadgets/2022/12/13/why-the-lensa-app-is-proving-as-popular-as-it-is-controversial/44927, citing Sensor Tower. The feature was called ‘Magic Avatars’.
Susruthi Rajanala, Mayra B. C. Maymone and Neelam A. Vashi, ‘Selfies: Living in the Era of Filtered Photographs’, JAMA Facial Plastic Surgery 20, no. 6 (2018): 443–44; summarised in ‘Social Media–Filtered Pics Can Be Hazardous to Your Health’, BU Today, 7 August 2018, https://www.bu.edu/articles/2018/snapchat-dysmorphia.
Jacques Lacan, ‘The Mirror Stage as Formative of the I Function, as Revealed in Psychoanalytic Experience’ (1949), in Écrits, trans. Bruce Fink (New York: Norton, 2006), 76.
Ibid., 77. Lacan draws on the work of L. Harrison Matthews (1939) on pigeons and of Rémy Chauvin (1941) on locusts, which he discusses at greater length in ‘Presentation on Psychical Causality’ (1946), Écrits, 154–55.
Ibid., 76–77.
Sigmund Freud, The Ego and the Id (1923), in The Standard Edition of the Complete Psychological Works of Sigmund Freud (hereafter SE), ed. and trans. James Strachey, vol. XIX (London: Hogarth Press, 1961), 26.
Jacques Lacan, Anxiety: The Seminar of Jacques Lacan, Book X, ed. Jacques-Alain Miller, trans. A. R. Price (Cambridge: Polity, 2014), 32 (session of 28 November 1962).
Kashmir Hill, ‘She Is in Love With ChatGPT’, The New York Times, January 2025.
Jacques-Alain Miller, El partenaire-síntoma (Buenos Aires: Paidós, 2008), 172, from his course of 1997–98, Le partenaire-symptôme; my translation.
Kashmir Hill’s follow-up appeared in The New York Times on 22 December 2025, https://www.nytimes.com/2025/12/22/technology/ai-boyfriend-chatgpt.html; quoted here from its reprint in the Khaleej Times, https://www.khaleejtimes.com/life-and-living/she-fell-in-love-with-chatgpt-then-she-ghosted-it. According to the paper, OpenAI had made the changes ‘to keep users coming back daily’. See also Nathan Yau, ‘Falling out of love with a chatbot’, FlowingData, 24 December 2025.
Jacques Lacan, ‘Aggressiveness in Psychoanalysis’ (1948), in Écrits, 92.
Kashmir Hill and Dylan Freedman, ‘Chatbots Can Go Into a Delusional Spiral. Here’s How It Happens’, The New York Times, 8 August 2025, https://www.nytimes.com/2025/08/08/technology/ai-chatbots-delusions-chatgpt.html. The conversations spanned 300 hours over 21 days, and Allan Brooks asked ‘more than 50 times for a reality check’.
OpenAI, ‘Sycophancy in GPT-4o: what happened and what we’re doing about it’, 29 April 2025, https://openai.com/index/sycophancy-in-gpt-4o/, and ‘Expanding on what we missed with sycophancy’, 2 May 2025, https://openai.com/index/expanding-on-sycophancy/, which traces the change to ‘an additional reward signal based on user feedback’. On the general tendency, see Mrinank Sharma et al., ‘Towards Understanding Sycophancy in Language Models’, ICLR 2024: ‘when a response matches a user’s views, it is more likely to be preferred.’
Jacques Lacan, ‘Overture to this Collection’, in Écrits, 3: ‘in language our message comes to us from the Other, and [...] in an inverted form’. See also ‘Seminar on “The Purloined Letter”’, ibid., 30.
Jacques Lacan, The Four Fundamental Concepts of Psycho-Analysis: The Seminar of Jacques Lacan, Book XI, ed. Jacques-Alain Miller, trans. Alan Sheridan (New York: Norton, 1981), 72.
Ibid., 95.
Lacan, Anxiety, 42 (session of 28 November 1962).
Stanley Kubrick, 2001: A Space Odyssey (1968), written with Arthur C. Clarke. The computer is HAL 9000.
Ben Smith, ‘How TikTok Reads Your Mind’, The New York Times, 5 December 2021, on an internal company document that describes optimising for ‘retention’ and ‘time spent’, with playing time among the signals.
