The Case I Put Into the Chatbot

The case

The Rev. David Kim is the sole pastor of a small congregation and has no regular supervision group. After difficult pastoral visits, he often feels uncertain about what to say next and has few colleagues available in the moment. He begins using a general-purpose AI chatbot as a reflection partner. He tells it to act like an experienced pastoral supervisor, summarizes the encounter, and asks for possible questions, prayers, and referrals.

At first David uses invented situations. Over time he supplies more detail because the responses become more useful when the context is specific. He removes names but includes ages, diagnoses, family relationships, immigration concerns, quotations, and recent congregational events. He sometimes pastes portions of emails so the chatbot can draft a compassionate reply. He has not reviewed the vendor’s data settings and uses the same personal account for travel planning and other tasks.

Following a visit with Grace, a longtime member dealing with her adult son’s addiction and a recent overdose, David asks the chatbot to identify the “real issue” and suggest a pastoral plan. The tool confidently characterizes Grace as codependent and recommends a firm boundary-setting conversation. David uses several of its phrases in a follow-up email. Grace experiences the message as clinical and blaming. She asks how he reached those conclusions after one visit.

David tells the wardens that he used an AI tool to help organize his thoughts. One warden is alarmed because the prompt included enough detail to identify Grace locally; another says this is no different from consulting a book or colleague. David feels ashamed and isolated. He did not intend to outsource care—he wanted help being a better pastor. The church has no policy, Grace does not yet know what information was entered, and David must decide what honesty, repair, and a safer consultation practice require.

What is known—and what is not

Known so far

• David entered sensitive pastoral information into a general-purpose AI service without the person’s knowledge.

• Removing a name did not remove distinctive, locally identifying details.

• The chatbot generated a confident interpretation that exceeded the available evidence.

• David incorporated its framing into pastoral communication and Grace experienced harm.

• His isolation and lack of human consultation are important conditions, not excuses.

Still uncertain

• What the service stored, retained, or used and whether the prompt can be deleted.

• Exactly what information David entered in this and previous consultations.

• What confidentiality, professional, denominational, insurance, or legal duties apply.

• What Grace would need in order to understand and respond to the breach of trust.

• What sustainable human supervision or peer consultation is available to David.

Pastoral response questions

Pause before solving. Imagine that you are the leader receiving this person or community. What do you notice, what do you need to learn, and what is the next faithful step?

Locate the pastoral task

1. What was David actually seeking: ideas, reassurance, supervision, diagnosis, language, or relief from isolation?

2. Which parts of pastoral consultation require an accountable human relationship?

3. How might a chatbot’s confident language affect a pastor who already feels uncertain?

Examine the use

1. What counts as genuinely de-identified information in a small community?

2. Which information should never be entered into a public or unapproved tool?

3. How should generated interpretations be treated when the system cannot examine, diagnose, or assume responsibility?

Consider people, power, and trust

1. What would Grace reasonably expect when speaking confidentially with her pastor?

2. Who benefits from this use, and who carries the risk?

3. How do pastoral authority and the asymmetry of the care relationship affect consent?

Discern boundaries and accountability

1. What should David tell Grace, church leaders, or others—and in what order?

2. What technical, privacy, pastoral, or denominational consultation is needed now?

3. How can accountability avoid isolating David further or reducing the matter to individual blame?

Plan faithful practice

1. What could a safe case-consultation template permit, caution against, and prohibit?

2. What human support structure should surround clergy using any reflection aid?

3. How should pastors document, disclose, verify, and take responsibility for AI-assisted pastoral communication?

Designing a tool for when you are stuck

Use these final questions to test what a practical pastoral field guide would need to provide.

1. What must a tool ask before any pastoral information is entered into AI?

2. How can it address re-identification, consent, vendor settings, and the limits of generated interpretation?

3. What alternatives should it offer a caregiver who is genuinely alone and stuck?

Developed from the AI Pastoral Toolkit Design Sprint case corpus and Day One Synthesis and Design Charge (July 2026).