Blog article AI and Religion: When Technology Becomes Artificial Religion

AI has long since become a religion – and not because people worship machines. The ways in which Artificial Intelligence (AI) is discussed and engaged with display structures typical of religions. When Sam Altman, CEO of OpenAI, describes AI as a “magical intelligence in the sky” (Murgia 2023), or when Demis Hassabis, CEO of Google DeepMind, outlines it as a solution to central societal problems (Perigo 2025), the status of these technologies shifts from that of a tool to that of an instance superior to humanity.

Following Mustafa Suleyman, currently CEO of Microsoft AI, AI can be described as a procedure for teaching machines capabilities that were previously considered the preserve of human beings. What is currently referred to as AI primarily comprises forms of Artificial Narrow Intelligence (ANI): technologies that, when given human instruction, can perform specific tasks – such as generating text or recognizing images (Suleyman and Bhaskar 2023). So-called Agentic AI has also emerged. These are systems that do not merely react to inputs but, within a limited frame, autonomously pursue goals and take action.

Artificial General Intelligence (AGI) must be distinguished from both Artificial Narrow Intelligence and Agentic AI: systems with generalized intellectual capabilities equivalent to those of human beings. Such an AGI does not yet exist. Nevertheless, it is effective – not as a technological reality, but as a future imaginary that shapes current discourse on AI. Statements made by leading actors in the tech industry, science-fiction narratives, and transhumanist visions of the future influence one another. Science fiction transports images of superior machine intelligence and autonomous systems. Transhumanism translates such images into a program of technological self-transcendence: human limitations are to be overcome through technological development, particularly in the areas of disease, aging, and death. In this context, the idea of a technological singularity gains plausibility. This refers to a hypothetical point in time when AI develops its intellectual capacities autonomously to the point that it escapes human control and predictability.

Narratives of this kind form the background of statements made by leading actors in the tech industry. Rather than restricting themselves to describing current AI systems, they project visions of the future that attribute to AI in general a particular problem-solving potential and a superior efficacy. For example, AI systems are expected to surpass the limits of human knowledge. The corresponding narratives go so far as to suggest that AI might inaugurate a new evolutionary stage for humanity. At this point, the religious semantics of such narrative patterns come into view: they extend far beyond the empirically verifiable properties of AI technologies. In these narratives lies the starting point of what I term “Artificial Religion”.

Before turning to Artificial Religion, a distinction is necessary: between developments in the fields of “AI and Religion” and “AI as Religion”. Under AI and Religion, two areas can be grouped. The first is the use of AI within existing religious contexts. Religious institutions employ chatbots to answer questions of faith, deploy avatar bots for catechetical purposes, or prepare sermons using Large Language Models (LLMs). Prayer bots generate individualized prayers and relocate religious practice into a digital interface. The second is the theological engagement with AI. Representatives of various religions ask what it means for machines to learn and to act within a limited frame, and they discuss questions of consciousness, responsibility, and the conception of the human. In both cases, AI remains tied to specific religions that have existed longer than AI itself: either as a tool of religious practice or as an object of religious interpretation.

The perspective of AI as Religion differs from this. Here, the focus is not on the use of AI within existing religions, but on whether attributions analogous to religion emerge in engagement with AI itself. A precondition for this conceptual move is clarity about what is meant by religion. Religion is heuristically defined here as a cultural practice in which assumptions that go beyond empirically verifiable causal relations become effective and are rendered plausible to people through material forms and practices. On this basis, the following argument can be made: in the engagement with Artificial Intelligence, in the attributions made to it, and in the practices associated with it, a new formation is taking shape, which I heuristically name “Artificial Religion”.

This Artificial Religion emerges through the interplay of AI narratives, stagings, and everyday experiences with AI. As outlined above, the AI narratives are charged with religious semantics, in which Artificial Intelligence appears as transformative and superhuman. Such attributions gain plausibility through repeated visual representations and everyday practices. A visual example is the so-called “AI Creation Meme”: a human and a mechanical hand reach towards each other, alluding to Michelangelo’s scene of creation. The motif appears on book covers and corporate websites, presenting AI as a creative force (Singler 2020). Comparable patterns can be observed in museums and future-oriented exhibitions that place AI at their center. Visitors move through darkened rooms in which luminous networks and large-scale projections dominate perception. While the technological processes that constitute the functionality of AI systems remain invisible, the impression is conveyed of a force that brings forth the future and shapes worlds. Instead of explaining AI in accessible terms, such technologies are rendered sensorily accessible as shaping instances credited with a problem-solving competence superior to human reason (Prohl 2025).

In everyday life, too, attributions of particular powers to AI systems consolidate. Millions of people use language models such as ChatGPT every day to draft texts, work through conflicts, or prepare decisions. They experience the systems responding within seconds. These responses generally appear coherent, factual, and thematically apt. Through the repetition of such interactions, the impression stabilizes that AI systems possess extraordinary competence – up to and including attributions of omniscience. Because most people carry their phones with them at all times – at the kitchen table or in the car – it becomes possible to encounter these instances, perceived as omniscient, at any moment. From this, an impression of omnipresence can additionally emerge. AI is thus available at all times, responds in a personalized form, and reliably provides an answer regardless of where one happens to be.

The particular problem-solving competence and efficacy assumed to inhere in AI systems become experientially accessible for many users, especially in their engagement with LLMs. This experience is not confined to language models, however. It points to a broader field of AI systems, apps, interfaces, and algorithmic infrastructures in which AI appears as an agentic and orientation-providing instance. Public AI narratives that attribute particular capabilities to these technologies thereby gain further plausibility. Whether as a promise of salvation or as an apocalyptic scenario, AI appears in both cases as a force whose possibilities extend far beyond ordinary technology.

Properties and capacities such as particular efficacy, omniscience, omnipresence, and superhuman competence have hitherto been attributed primarily to goddesses and gods or to other instances understood as transcendent. In classical religious constellations, such assumptions gain plausibility through bodies, spaces, images, sounds, and situational arrangements. The meaning attributed to them is determined by the interpretive practices of those involved. In the engagement with AI, comparable attributions shift onto a technologically generated counterpart that is based on algorithmic processes and responds immediately.

This alters the configuration of mediation through which religious relations are established. In classical religious practices such as prayer, religious actors address themselves to goddesses, gods, or other instances located beyond the human. They formulate concerns, ask for orientation, or expect guiding signs. Whether and how an answer is experienced depends on the interpretation of those praying. They may relate events, feelings, later developments, or experiences within the ritual to the addressed instance and understand them as a response. Meaning here emerges through the interplay of practice, expectation, and interpretation.

In the engagement with LLMs, this relation shifts. What in religious constellations must be interpreted by practitioners is here delivered as a linguistic reaction directly by the system. To respond, in the case of LLMs, means that the systems generate a personalized, context-sensitive, and semantically apt reaction. Particularly with personal questions or interpretations of the future, users report that LLM answers are so precisely targeted that they seem guided by a power superior to humans. What recedes into the background is that this responsiveness is technologically generated and reproducible. Attributions of particular efficacy, which emerge from the interplay of tech narratives, transhumanist visions of the future, popular culture, and science fiction, thus do not remain confined to narratives or expectations. They become experiential in concrete interactions and are repeatedly rendered plausible in their performance. In this way, algorithmic authority emerges: AI not only provides suggestions but also takes over functions in everyday life comparable to those of religious formations. It offers orientation and shapes expectations of the future.

Users generally do not understand their interactions with AI as a religious practice, nor do they refer to them as religion. For analytical purposes, this self-designation is not decisive. What is central is the question of which attributions become effective, how they stabilize, and how they shape action. Applying the concept of religion to the novel cultural phenomenon of social interaction with AI is a matter of making visible how supra-ordinate authority is attributed to a technologically generated instance, how this attribution gains stability through repeated practice, and what consequences it has in everyday life.

Religion is understood here fundamentally as a cultural product: it takes shape where assumptions that exceed the empirically verifiable are rendered plausible and stabilized through practices, interpretations, and medial arrangements. Herein lies a first point of comparability with AI, which likewise emerges as a cultural product shaped by discursive, institutional, and technological conditions. At the same time, religious formations are materially and technologically mediated – through prayer wheels, church spaces, sound, light, images, or elaborately framed ceremonies. In the case of AI, a new quality of this mediation is added: algorithmically generated responsiveness. The instance to which particular efficacy is attributed here is, on the one hand, culturally produced and, on the other, technologically realized; it generates answers that appear personalized, situational, and in real time. In its performance, it is thus experienced as all-encompassing, addressable at any moment, and superior to human beings. People have AI models draft texts and prepare decisions because they interpret their answers as authoritative. Through such attributions, AI practice becomes comparable to religious practice. What was previously merely a religious attribution here becomes a technologically realized performance.

We refer to the alleged intelligence of the machine as artificial – following this, I call the religion that emerges from the machine “Artificial Religion”.