Report
A three-year theological inquiry reaches its scheduled start. Earlier research helps explain why feeling understood and participating in a relationship are different questions.
A KU Leuven theology project enters its scheduled research period on October 1 with a question directly relevant to churches adopting conversational AI: how does dialogue with a simulated interlocutor shape religious identity? The university lists “Artificial Identities? Navigating Agency and Faith Identity Construction in the Age of AI” as running until September 30, 2029. Its public entry describes an investigation to be undertaken, rather than completed findings about believers. Source: research.kuleuven.be
The entry names Ryan Kevin Mc Aleer as the fellow, Lieven Boeve as promotor and Didier Pollefeyt as copromotor, with support under an FWO junior postdoctoral program. The proposed approach brings philosophical accounts of selfhood into conversation with theological anthropology. It asks how to distinguish a genuine dialogical relationship from algorithmic imitation, while considering ways AI might also assist human dialogue.
No new experimental result, participant sample or validated assessment tool is reported in that project description. The immediate development is the scheduled beginning of a three-year inquiry. Its premise should not be mistaken for evidence that a particular chatbot has already changed a particular community's faith.
A useful comparison comes from an earlier experiment about feeling understood. In a 2024 PNAS paper, Yidan Yin, Nan Jia and Cheryl Wakslak separated two questions: who actually wrote a supportive response, and who the recipient was told had written it. Participants described difficult situations; responses came from another participant or Bing Chat. The researchers then varied the label identifying the apparent source. Source: pmc.ncbi.nlm.nih.gov
The main analysis included 455 responses. Participants were recruited through Prolific with screening for US residence and English as a native language. AI-written messages produced stronger feelings of being heard, but an AI label reduced those feelings. Message quality and beliefs about the responder therefore contributed separately to the experience.
This was not a study of churches, conversions or long-term spiritual formation. It does not establish that AI possesses empathy, nor that a supportive response creates an enduring relationship. It does show why a theological inquiry cannot simply dismiss the user's experience because the response was machine-generated. The experience and its interpretation require different kinds of examination.
The question also has a constructive side. Jeff Clyde G. Corpuz's March 12 article in Religions develops a practical-theological framework for AI use in interreligious dialogue. It considers translation, textual analysis and comparison across religious sources, while keeping institutional responsibility and human agency central. The framework is a scholarly proposal, not proof that adopting it improves relations between communities. Source: mdpi.com
Corpuz warns that automated interpretation can flatten theological differences or overlook practices that are not adequately represented by written text. That concern matters when an assistant summarizes what another religion believes. A fluent answer may conceal the difference between describing a tradition's literature and encountering people who practice it.
Taken alongside the KU Leuven inquiry, the paper suggests two questions that should remain separate. Does software help people prepare for dialogue across a language or knowledge gap? And does it encourage them to substitute generated exchanges for encounters in which another person can correct their assumptions? These are questions for evaluation, not reported conclusions from either project.
A congregation does not need to settle every philosophical question about AI to describe its own use clearly. Suppose a study group uses an assistant to generate questions before meeting. That hypothetical workflow can be evaluated by checking the questions against the assigned material and observing whether participants still engage with one another. Calling the assistant a spiritual companion introduces a different expectation, one that accuracy checks alone cannot resolve.
The difference is practical. A person can return to a teacher, explain how guidance affected them and ask the teacher to take responsibility for a misunderstanding. An interface may produce another response, but the institution offering it still needs to identify who can investigate a complaint or correct the underlying material. These are editorial implications of the comparison, not requirements already established by the newly listed project.
Such distinctions also leave room for beneficial use. Translation support, accessible explanations and preparation for a difficult conversation can serve human relationships. The relevant test is what the tool enables people to do together and what responsibilities remain visible, rather than whether every digital exchange is intrinsically helpful or harmful.
James 1:19–27 connects readiness to listen with restraint in speech, action on what is heard and care for vulnerable people. The passage does not treat religious understanding as the successful reception of agreeable words. It asks what follows from hearing. Read the passage in the NIV: James 1:19-27 (NIV)
For Christians considering conversational AI, that context provides a useful question: does this exchange prepare us to listen more carefully and act more responsibly toward real neighbors? The application is theological, not a claim that James anticipated language models or that a Bible passage measures software performance.
A response can be reassuring while leaving those obligations untouched. Equally, a helpful explanation can become part of genuine learning when a community examines it and puts sound understanding into practice. The work now scheduled at KU Leuven places the relationship behind the response under scrutiny. Its eventual value will depend on the clarity of the distinctions and evidence it produces, not simply on the urgency of the question it asks.
KU Leuven research portal: Artificial Identities?
Primary university project entry inspected; scheduled period and aims reported, no implementation or findings claimed. Independently authored 2024 experiment and 2026 theological paper supply dated context. No interviews, participant reactions or clinical conclusions.
News brief / Education & families
An October 5 ministry account gives a cumulative participation total across 21 districts and cities, while keeping lesson verification and religious formation with teachers.
Published 2026-10-08 · Simeulue event and ministry report October 5, 2026; cumulative training period unspecified; UNESCO background 2024
News brief / Education & families
The October 8–10 programme connects Christian teaching with classroom choices. Earlier student-writing research adds context, with clear limits on what it represents.
Published 2026-10-08 · Conference scheduled October 8–10, 2026; Chimes report October 5; research background July 22
News brief / Education & families
Fordham’s conference follow-up calls for work across institutions. IAJU’s resource plans outline a pilot hub and regular exchanges, with delivery still to be established.
Published 2026-10-08 · Conference October 1–3, 2026; Fordham report October 6; undated IAJU plans checked October 8