For George Davila Durendal’s separate 2020 text-generation and portrait-animation experiment, see AI_Jesus (Durendal, 2020).
| Alternative names | AI Jesus; Deus in Machina; chatbot Jesus; Jesus chatbot; AI confessional; AI Jesus avatar |
|---|---|
| Type | Machine-mediated religious interaction case / public religious-AI experiment |
| Hieropedia status | Documented public case; broader category still emerging |
| Field | Machine hierology |
| Documented setting | Peter's Chapel (Peterskapelle), Lucerne, Switzerland |
| Primary experiment | Deus in Machina, August–October 2024 |
| Interface | Confessional-style booth with Jesus avatar |
| Technical stack | GPT-4o, speech recognition, and avatar generation reported |
| Human involvement | Researchers tested the system; church personnel supported visitors, with a theologian often present for further pastoral dialogue[8] |
| Current status | The 2024 Lucerne installation was temporary; HSLU lists the 2025–2026 AI-Jesus 2.0 research phase as completed[5] |
Deus in Machina, often called AI Jesus, was a temporary installation at Peter’s Chapel in Lucerne, Switzerland, from August to October 2024. Visitors spoke with a Jesus avatar inside a confessional-style booth. The organizers presented the encounter as an experiment in religious conversation and explicitly excluded confession and absolution.[1][2]
A chapel encounter
The installation joined three elements: a represented sacred figure, software producing a reply, and a chapel setting that shaped expectations of the exchange. Visitors faced a long-haired avatar behind a lattice screen; speech recognition, GPT-4o and avatar software supported conversations in several languages.[1][2] The setting made the project distinct from a general scripture-search application.
The 2024 installation
Project researcher Philipp Haslbauer’s report of 25 November 2024 records around 900 conversations during the exhibition. That is a count of conversations, not unique visitors.[7] Researchers tested the system, chapel personnel supported visitors, and a theologian was often available for further pastoral dialogue. Human support surrounded the encounter without determining every generated answer.[8]
Prompting and human involvement
A 2025 author preprint reproduces the base prompt and describes Bible retrieval, conversational memory and turn-specific directives. These components shaped the avatar’s role and the course of the exchange, including its closing prayer or blessing. The paper also describes an information message, a consent button, warnings against identifying disclosures and filtering of transcribed input. It is a first-party technical account: it makes part of the prompt and implementation inspectable, but does not supply a complete runtime record or independent safety validation.[8]
The authors distinguish iterative testing from control over the exact replies. That distinction matters to the chapel’s pastoral role: a human-designed setting and an available theologian did not make each answer a statement approved by the church.[8]
Reception and criticism
The Guardian reported spiritually meaningful visitor feedback alongside people saying that they could not talk to a machine. It also described a local journalist’s criticism of repetitive answers and theologian Marco Schmid’s assessment that some responses were superficial. Schmid opposed permanent installation because an inappropriate answer could not be ruled out.[1]
AP reported organizers’ favorable impressions of visitors leaving moved or thoughtful. It also quoted AI and Faith writer Kenneth Cukier on possible connection and on the risk of distancing people from deeper spirituality. These accounts document reactions, rather than independently measured pastoral benefit.[2]
The objections concerned both the representation and its placement. The Guardian reported objections from some Catholic colleagues to the confessional setting and from some Protestant colleagues to the imagery. Haslbauer told AP that online commenters had called the project blasphemous or the work of the devil.[1][2] People’s later story is additional circulation of this controversy, rather than a separate technical assessment.[3]
AI-Jesus 2.0
Lucerne University of Applied Sciences and Arts (HSLU) lists a successor project, AI-Jesus 2.0, running from 1 September 2025 to 30 June 2026. The record describes spiritual-conversation concepts, platform design and implementation partners, and marks that research phase as completed.[5]
A partner invitation of 6 March 2026 described a working avatar prototype and sought collaborators on dialogue about meaning, values and orientation. Web and mobile services and additional personas were proposed directions for development. The materials establish a prototype and research programme; they do not verify a public launch of those services or their pastoral effects.[6]
Religious mediation and related tools
The case documents a church-hosted encounter with an AI representation of Jesus. Researchers and chapel staff staged it within an existing Christian setting; the record does not establish a new doctrinal institution.[8] Le Monde places it beside CatholicChat, AskBuddha and Rabbi Bot. The comparison connects the installation to religious conversational interfaces, while its confessional architecture and public chapel presentation remain distinctive.[4]
References
- The Guardian, “Deus in machina: Swiss church installs AI-powered Jesus.” 21 November 2024.
- Associated Press, “'AI Jesus' avatar tests man's faith in machines and the divine.” 28 November 2024.
- People, “'AI Jesus' Installed in Church's Confessional Booth Dubbed 'the Work of the Devil' by Critics.” 5 December 2024.
- Le Monde, “Comment l'intelligence artificielle bouscule les religions: ‘AI Jesus’, ‘Ask Buddha’, ‘Rabbi Bot’...” 9 February 2025.
- Lucerne University of Applied Sciences and Arts, “AI-Jesus 2.0 – Immersive AI Agents for Spriritual Conversation.” Project record, period 1 September 2025–30 June 2026; status consulted 3 October 2026.
- HSLU, “Wenn KI zuhört: HSLU-Forschungsteam sucht Partner für KI-Jesus 2.0.” 6 March 2026.
- Philipp Haslbauer, “Experimental Art Installation ‘Deus in Machina’ – First Results.” 25 November 2024, pp. 1–2. Organizer research report.
- Philipp Haslbauer, Carolin Reichherzer, Marco Schmid and Aljosa Smolic, “Deus in Machina: A Conversational AI for Spiritual Experience.” Author preprint internally dated October 2025, 10 pages; cited sections 3, 4.2, 4.4–4.5 and 5.2. First-party technical account.
Public reporting documents the Lucerne experiment and attributed reception; HSLU materials describe successor research, and the author preprint supplies first-party implementation detail. The cited preprint is distinct from the publisher’s final text, which was not inspected. These sources do not establish sacramental validity, divine agency, permanent clerical replacement, complete prompt/runtime context, or a complete technical or safety audit.