Singapore theologian distinguishes AI study tools from divine authority

A newly published theological argument

Singapore theologian Roland Chia has warned Christians against treating chatbot language as communication from God. His November 2 essay for the ETHOS Institute for Public Christianity distinguishes using software to explore religious information from giving it spiritual authority. It is a newly published theological argument, not a new study of chatbot users or the launch of a religious app. Source: ethosinstitute.sg

Chia, a Methodist and professor of Christian doctrine at Trinity Theological College, argues that apparently confident answers can be mistaken for revelation. His concern includes dependence on a simulated religious figure and the displacement of prayer, theological reflection and human pastoral presence. Those are his theological judgments and cautions, not measured outcomes established by the essay. Source: ttc.edu.sg Source: ethosinstitute.sg

What the software maker says

There is an important distinction between that concern and what a developer officially promises. Text With Jesus, one of the services Chia names, says in its public FAQ that its responses are language-model output. It describes the app as a means of exploring biblical narratives and expressly disclaims actual divine insight or consciousness. The FAQ does not describe the service as direct communication with God. Source: textwith.me

That is relevant countercontext, not proof that every user understands the distinction. We inspected the public explanation, not a logged-in conversation, and did not test whether every generated reply maintains the stated boundary. Nor did we establish that users experience the harms Chia fears. His essay cites older examples and reporting; those should not be mistaken for newly observed November incidents.

The resulting question for a church educator is narrower and more useful than whether a chatbot sounds religious: can the class distinguish a quotation from Scripture, an author's interpretation and newly generated speech in a biblical character's voice?

A boundary readers can apply

For example, in a hypothetical study session, participants could mark those three kinds of text separately and check any claimed quotation in its surrounding passage. A first-person voice should not erase the distinction. Someone who chooses not to use a character simulation can examine the developer's public description without creating an account or sharing personal concerns with it.

For Christians, Hebrews 1:1–4 (NIV) identifies God's Son in relation to creation, revelation and purification for sins. Its claim concerns who Christ is, not simply the persuasiveness of words attributed to him. Applying that distinction to software is Christian interpretation, not a scientific finding about a model. Hebrews 1:1-4 (NIV)

A disclaimer can identify a simulation's intended limits. It cannot, by itself, supply pastoral accountability or settle the authority a person grants its answers. Reading both the theologian's argument and the developer's stated boundary helps keep those separate questions visible.