An evangelical argument for keeping AI in a supporting role

The starting point

Willy Kotiuga’s November 21, 2025 Lausanne article argues that AI can assist communication and analysis while remaining subordinate to human discernment and spiritual responsibility. Its editorial note discloses AI assistance in research and editing. This is an identifiable evangelical contribution, not a position survey across Christian traditions.

Why the servant metaphor helps—and where it stops

The interesting test is whether an organization can describe what it refuses to delegate. “Human oversight” remains abstract until someone names the decision they will personally examine and accept responsibility for.

A ministry might use a system to reorganize notes while retaining responsibility for the message’s accuracy, pastoral appropriateness and theological claims. That distinction can be lost when an efficient workflow quietly becomes a habit of accepting whatever sounds convincing.

Calling AI a servant gives Christians a useful starting point: the tool should support human purposes rather than determine them. But the metaphor does not settle which purposes are good or which tasks can responsibly be delegated. A system used to save time might still undermine a relationship if the time-saving measure removes the attention that relationship requires. The test is not simply whether a human gives the instruction.

Lausanne's contribution belongs to an evangelical conversation about mission and discernment. It is a situated Christian argument, not a statement establishing what all churches believe. Readers should engage its reasoning without treating the institutional platform as proof of universal agreement. Its disclosed use of AI assistance is also relevant: the piece participates in the practice it discusses. Disclosure helps readers understand its production, but the strength of its claims still depends on the argument and evidence.

Assistance can change the work itself

Consider sermon preparation. Software might help a minister locate a passage, organize notes or compare possible outlines. Those tasks can be useful, but an outline is not the same as understanding a congregation's needs or taking responsibility for what will be proclaimed. If the minister accepts an interpretation because the wording is polished, the tool has moved from assisting preparation toward supplying unexamined authority. This is a hypothetical distinction, not an accusation against a particular preacher.

The same issue appears in pastoral correspondence. A suggested article can help someone who struggles to find words. Yet a condolence message sent without attentive reading can misrepresent the sender's care or include an inappropriate assurance. The recipient encounters the message as part of a human relationship. The person who sends it remains responsible for its truthfulness, sensitivity and promises, regardless of how the article was produced.

A church policy therefore needs to describe tasks rather than merely declare itself for or against AI. Administrative sorting, public teaching, confidential counseling and decisions affecting employment have different stakes. Calling all of them assistance can hide those differences. Calling all of them replacement can be equally unhelpful, because it overlooks limited uses that preserve responsibility and improve access.

Discernment needs more than a feeling of reassurance

Spiritual language can encourage careful judgment, but it can also become vague. A leader may sincerely feel at peace about a tool while lacking information about its data practices or accuracy. Another may feel anxious because a technology is unfamiliar. Neither feeling alone supplies the evidence needed for an institutional decision. Discernment involves moral reflection together with attention to the actual circumstances.

For example, before using a service to summarize sensitive conversations, a ministry should know what information is uploaded, who can access it and whether participants consented to that use. These are questions to investigate, not claims about the behavior of any named service. A prayerful intention does not answer them. Conversely, knowing the technical details does not determine whether the use respects the person whose story is involved.

The combination matters. Christian institutions should resist separating spiritual reflection from practical accountability, as though one committee handles faith while another handles consequences. A policy is more credible when the people affected can understand it and challenge its application. The language of stewardship becomes concrete when someone is identifiable as responsible for a decision and capable of changing it.

The risk of substituting output for formation

A student can produce a plausible theological paragraph without learning to interpret a text. A ministry team can produce an impressive strategy without agreeing on its priorities. AI does not create every version of this problem, but inexpensive polished output makes the distinction between completing a document and doing the underlying work especially important.

Formation involves habits: reading patiently, listening to another person, testing an assumption and accepting correction. These cannot be measured only by the quantity of finished material. A church using AI in education should ask which habits an assignment is meant to cultivate. Assistance that removes an irrelevant obstacle may help; assistance that performs the central act of learning may defeat the purpose.

There is a fair counterargument. People also learn through examples, dictionaries, editors and conversation partners. Requiring every task to be performed unaided would be an arbitrary standard. The better question is whether the learner can explain, evaluate and take responsibility for the resulting work. A tool can support that ability or conceal its absence. The teaching design, not the mere presence of software, makes the difference.

Christian perspective: wisdom is embodied in conduct

James 3:13–18 describes wisdom through humility and the character of a person's conduct. In its setting, the passage concerns life in a community, including the destructive effects of rivalry. It is not a prediction about artificial intelligence. Its relevance here is that impressive speech is not sufficient evidence of wisdom. The way knowledge is used, and the relationships it serves, matter.

Applied to ministry, this means that an eloquent generated answer should not be confused with a wise pastoral response. Wisdom includes knowing when to speak, when to listen and when to seek help from someone with the appropriate responsibility or expertise. A system's ability to formulate a sentence does not establish that it bears those obligations or understands the person before it.

This perspective also disciplines human leaders. It would be inconsistent to criticize machines for lacking accountability while refusing questions about one's own use of them. A minister who uses AI should be able to explain the boundaries of that use without claiming that personal status makes scrutiny unnecessary. Service is strengthened by answerability, not diminished by it.

In 1 Thessalonians 5:21, the instruction to test and hold fast to what is good belongs to the life of a discerning community. It encourages examination rather than either blanket suspicion or passive acceptance. Applying it to AI does not make a generated response a prophecy.

The practical Christian question is whether assistance strengthens responsible judgment or allows its avoidance. A polished sermon outline cannot know the congregation, repair a broken relationship or bear the cost of an untrue accusation. Those responsibilities remain with people.

A workable ministry conversation

A useful staff discussion starts with one real task. Describe what assistance is proposed, what information it needs, who will check the result and what happens if it is wrong. Then ask what part of the task must remain a human responsibility. For a public Bible study, that might include verifying references and owning the interpretation. For a confidential conversation, it may mean declining to upload the material at all unless the institution has established a suitable process.

Record a limited decision and revisit it after experience. Do not turn a successful use in one setting into blanket permission for every other setting. Equally, do not let a poor result in one task settle the value of all possible applications. Such distinctions make the servant metaphor operational: the institution defines the purpose, limits the delegation and remains accountable for the outcome.

Lausanne's article offers a theological contribution to that conversation. Our assessment is that its practical value depends on translating its language into visible choices about attention, authority and responsibility. A Christian institution serves people more faithfully when it can explain those choices than when it merely adopts an inspiring description of technology.

Write down which uses are allowed, which require review and which decisions must remain personal. Explain the role of AI honestly when that role would affect a listener’s understanding of authorship or care.