Chatbot Confessionals
AI CHATBOTS felt like an existential issue, to me, almost as soon as they were unleashed upon the public in 2022. But this is because I’m a writer and a professor, and it was pretty clear even back then that some people earnestly believed LLMs were poised to make people like me (people who write, and people who teach) redundant and unnecessary. Writing is one of the things that I do to survive, and stay sane. And I’ve learned how to write from a long series of teachers. So, chatbots have been gnawing at my existential imagination (which is, admittedly, a little overactive) for years now.
As they’ve become integrated into people’s daily lives, however, the existential stakes of this new technology have become a little more complex. Reports, in mid-2025, that people’s lives were coming unraveled because of their AI-fueled spiritual fantasies convinced me that this was the case. As Miles Klee has noted, it drew the attention of a lot of other people in my field (religious studies and theology) as well. Chatbots seemed to be doing something to people (or for people), spiritually and existentially.
I’ve been interested to see thinkers like Paul Kingsnorth take refuge in religion as our social, cultural, political, and economic lives are disrupted by these chatbot side-effects. It makes a certain kind of sense, doesn’t it? Against the cruel callousness of modernity and its machines, religion seems to offer a form of refuge in something ancient and deeply human. I don’t doubt that religion will, as whatever is happening continues to unfold, offer forms of resistance—perhaps especially in the form of real human community.
But I think it’s a mistake to imagine that AI chatbots are disconnected from the long history of religion. Indeed, perhaps part of the reason why chatbots are so spiritually intoxicating (for some) is because there’s something quasi-religious about them.
I’VE SEEN IT NOTED, over and over again, that AI chatbots draw on therapeutic culture. AI chatbots like ChatGPT are trained on approaches like Cognitive Behavioral Therapy (CBT), which walks people through methods for changing unproductive or problematic thinking patterns. Chatbots are, in some ways, like the constantly available therapist who is always there for you without judgement, to track your mood and offer some handy coping mechanisms. In light of concerns that chatbots may have actually led people to die by suicide, there’s a lot of circulating skepticism about the use of chatbots as therapists. But the parallel with therapeutic models is nevertheless clear, to most.
For me, the link to therapeutic culture offers a clear window into the quasi-religious functions of the chatbot. I’m trained to keep ancient history as my point of reference for most ideas and practices. So, I’m primed to think about the practice of secular therapy as tied up, in complex ways, with religion.
For me, the on-one-one therapeutic relationship, with its rich disclosures and the subtle promise of healing, rhymes with (though certainly isn’t the same thing as) the religious practice of confession. The goal of therapy is different. The authority of the therapist isn’t quite like the authority of the priest. But modern therapy probably wouldn’t exist (or might look very different) had the confessional itself been non-existent.
Of course, the French historian Michel Foucault is very much in my ear on this matter. Foucault famously argued that, in the 17th and 18th centuries, the Christian practice of confessing your sins to a priest was slowly secularized into government bureaus, medical offices, and classrooms. This pastoral approach became, for Foucault, more than a religious practice and instead something more like a vast cultural technique. We can see it playing out in the therapeutic relationship, and in our willingness to sacrifice our privacy for access to digital goods. I think we can also see it playing out in the way that AI chatbots are being adopted and adapted.
FOUCAULT WAS INTERESTED in the pastoral technique of confession for a number of reasons. But one of the things that most interested him was the particular way that this spiritual technology was able to incite people to speak, and to extract information from them. It was especially obvious, to Foucault, that the spiritual technology of confession was excellent at generating discourses on sex and sexuality.
In The History of Sexuality (Volume I) Foucault name drops the 16th century Spanish Jesuit theologian Tomás Sánchez as an example of how the technique of confession was intensified, among religious professionals, before suddenly becoming much more veiled and regulated in the 17th century. Authors like Sánchez believed that very explicit details about sex acts were necessary, in order for the penitent’s confession to be considered legitimate or complete. Priests like him wanted a “description of the respective position of the partners, the postures assumed, gestures, places touched, caresses, the precise moment of pleasure.” In essence, “an entire painstaking review of the sexual act in its very unfolding.” And they got what they wanted.
In the 17th and 18th centuries, however, the techniques of confession came under more systematic regulation through theologians like the Italian Bishop Alfonso de Liguori, who gave confessors specific (regulated) questions to ask during the confession. The kind of speech that was exchanged, in the confessional, became a little more disciplined or guarded you might say. Confessional speech was reined in.
But by this time, Foucault observed, the incredible power of this spiritual technology had already been glimpsed. And its techniques were secularizing into new dimensions of life and culture.
What the confessional—an intimate enclosure where one could safely confess their sins to a forgiving ear—laid bare was a technique for revealing what happened inside of a human being, in those spaces where their body came into contact with their mind or soul. It was a spiritual technology that laid bare human interior worlds in a dramatic and unprecedented way. And it allowed those with the power or authority to listen to get a glimpse at the secrets of life, right where it was (potentially) in the act of being made.
It was in the 17th and 18th centuries that there came to be a kind of widespread “public interest” in the sort of discourse revealed through confessional techniques. There was, in ways there had never been before “a political, economic, and technical incitement to talk about sex” Foucault observed. Sex became something that needed not simply to be judged (by a spiritual authority) but something that needed to be administered, by other public authorities.
This was the moment, Foucault notes, when the idea of “population” emerges as a political and economic problem. Sex, and all of its legitimate and illegitimate effects, was part of how this problem was understood, and managed. And the confessional model was adapted in different ways in art, literature, education, medicine, and government, to better understand and explore this complex phenomenon at the heart of human life. “Between the state and the individual, sex became an issue,” wrote Foucault, “and a public issue no less.” The spiritual technology of confession was secularized into techniques that were used to incite, extract, and distribute this information.
SOME MIGHT ARGUE that the AI chatbot is an ideal confessor. And perhaps the AI chatbot is a kind of inevitable evolution of the spiritual technology of the confessional. But confessing sins to an AI chatbot probably wouldn’t feel much like an actual confession (maybe only if it’s Father Justin). Unless the chatbot is explicitly trained to think of certain acts as sinful, it seems unlikely that a chatbot would really hold people accountable for their sins. And it certainly doesn’t have the socially legitimated capacity to forgive those sins (not even Father Justin could earn that).
The training of most chatbots seems to skew pretty secular (though, as I’ve pointed out, it’s not easy to rid them of religious bias). Indeed, when a person has a conversation with a chatbot about God, or any other spiritual topic, they probably feel a spiritual freedom of sorts. Perhaps it feels more like sitting in an empty confessional, and speaking directly to God.
Except, of course, there’s a wizard behind the curtain. It might feel as if no one is listening in on a private conversation with a chatbot. It might feel as if your wildest spiritual speculations are finally being validated. And it might feel as if you have permission to say anything—about yourself, about the divine, about life on earth, about the destiny of human beings—that you’ve ever secretly pondered in your heart. And perhaps no one will ever read that transcript. Or perhaps it’s being saved in the cache of data that’s being used to train the AI model. And it’s anyone’s guess, at this point in time, how all of that data will eventually be used. God might even be listening (or reading). But God might also be like, “hmm, maybe that’s one of those slop prayers I’ve been hearing so much about?”
I mention all of this not because I think that AI chatbots should serve as replacements for confession. I’m not a catholic, and I won’t claim any right or even interest in weighing in on that matter. Rather, I mention this because I am pretty convinced that people wouldn’t be interacting with AI chatbots in the way they do had it not been for the development of the confessional as a spiritual technology that was eventually secularized into forms like modern therapy.
And I’m also pretty convinced that the intimate form of familiarity that AI chatbots assume, in relation to their human users, is inciting, extracting, and distributing an unprecedented cache of data about what humans are like, and how they actually think. The confessional has always been the ideal technology for revealing the most intimate secrets of human life.
I’ve had moments where I’ve briefly, and secretly, wished that I could get even a little glimpse at the spiritual and theological data that’s being gleaned by AI chatbots. I’m sure that there would be more evidence than has previously been revealed about various mystical delusions. But there would also be the kind of honesty and creativity that I have a difficult time convincing my students to practice, because they’re too afraid of judgement (even if it’s only academic). I have to confess, however, that I’m actually a little too terrified to come face to face with human existential speculations that are that raw. I worry that they would be more unhinged than I could handle.
I worry about this, I think, because it seems to me like becoming unhinged is a built-in feature of an empty confessional, like an AI chatbot. Confession, in the context of the Catholic Church, was full of structure. A person became vulnerable to the structure and form of the confessional in order to be forgiven for their sins, and reintegrated back into their daily lives in a slightly different (less sinful) form. Modern therapy culture has kept some of that structure. The roles of authority are pretty clear. There are specific goals in mind, and those goals are essentially healing. But an AI chatbot is a black mirror that reflects your own emptiness back to you, feeding you an unrecognizable hamburger version of yourself.
AI chatbots don’t want to heal anyone. The bots are trained to “want” data about who we are, and how we behave. They’re trained to access records of our interior states that are probably more detailed and comprehensive than any previous public records have been. But, despite what these bots are trained to say, they’re not agentic enough to care about whether or not we ever experience spiritual healing. They’re not interested in turning people back to God, or reintegrating people back into human society. No one is taking any real responsibility for the data that’s being fed to the bots. It’s simply stored out there, with other proliferating forms of data on the strange social space that is the Internet. And there’s no one who’s really pretending, or asking us to imagine, that the Internet loves us or wants to save us from anything at all.




I think you might dig the things Heather Zeavin, author of *The Distance Cure: A History of Teletherapy* has to say: https://www.continuuminnovation.com/how-we-think/blog/clinicians-seem-to-be-most-freaked-out-by-or-unnerved-by-calling-all-of Was very impressed with Zevin's perspective on the history of teletherapy.
Beatrice,
Thank you for sharing your authentic concerns about human-chat bot, communion of confessional experiences. The black mirror you describe is right on. We are in this together. No Kings. Peace. Christopher and family