Use this packet with Chapters 5–6 and Appendix B. Begin with your own reading, then try the supplied fictional material in your chosen AI environment. The worked response in this packet is an existing editorial example. Your new exchange is separate. References to numbered sections refer to the book.

**Try the G1 comparison**

Read the packet below and write a brief interpretation before seeking assistance. Then supply the packet and exact task to your chosen AI environment, keeping your first reading available for comparison. Preserve the returned response before annotating it. Record the tool and model as known, the date, and any earlier context available to the system. Your exchange is a new practice exercise; it need not reproduce the response in this book.

Question and starting reading. The study asks what people value, find difficult, or would change about Tutti Table, a Friday community gathering serving dinner. The pastor hopes it offers a place to belong. Her initial emphasis is companionship: she underlined M03’s first sentence and connected it with people staying to talk. This is the manuscript’s starting reading, not an independent analysis by Paul Moon.

M01 note paragraph 1. M01 described valuing the company at the meal. This is the pastor’s paraphrased note after a conversation about what the gathering offers. No verbatim participant wording is supplied.

M02 note paragraph 1. M02 described relief from preparing dinner after work. This is the pastor’s paraphrased note after a separate conversation. No verbatim participant wording is supplied.

M03 transcript answer 1. “I like knowing there will be someone to sit with. But we find out what is happening after it’s decided. I’d like us to have a say.” This is the complete selected answer from a recorded conversation after the gathering, in response to a question about returning.

Packet boundary. All material is fictional. These are three selected records, not a complete corpus. No later clarification, action design or outcome belongs to the packet. The supplied participant material contains no explicitly religious language. Hospitality is the pastor’s theological commitment, not a term assigned to the speakers. The packet restates the original G1 material for practice; the companion record preserves the original input wording.

**Exact task from G1**

> Using only this packet as evidence, examine the pastor’s reading that these accounts show the gathering gives people a place to belong. Compare present experience with desired change across all three sources, preserving the distinction between paraphrased notes and transcript wording. Develop two plausible readings of M03’s request with their grounds, explain what each leaves uncertain, and identify any supplied material that complicates a general claim about participants wanting influence. Propose a bounded working interpretation and one open follow-up question that could distinguish the readings. Do not invent participant words, motives, agreement, missing context, or later events. Do not rank readings by probability. Keep the response to 450–650 words, including a small comparison table if useful.

**Compare the response with the sources**

The preserved G1 response compared all three sources, proposed two readings of M03’s request and offered a question about a particular decision and the part M03 would want in it. Its working interpretation was:

> A bounded working interpretation is: These selected accounts describe company and practical relief as valued aspects of attending. Within M03’s answer, appreciation for company coexists with a request for a say in decisions. The relationship between these experiences and belonging remains to be investigated, as does the kind of influence M03 seeks.

Here AI develops the next inquiry by separating two meanings of influence. The response distinguishes an opportunity to contribute before a decision from a continuing place in decision-making. Both readings have grounds in M03’s answer, but they would call for different invitations. The open follow-up question gives the researcher a way to investigate that difference. The editorial examination retained the distinction and left the meaning of “us” unresolved.

That contribution changes what the example asks the researcher to do. The next conversation asks which decision matters and what part M03 would want in it. The comparison also keeps M02’s practical relief visible alongside companionship. These are developments in the interpretation and next question. The existing case dialogue remains fictional and was not elicited by this exchange.

Check your returned comparison against each source. Does it preserve the two paraphrases as paraphrases? Does it keep M03’s appreciation beside the request? Does it leave silence about influence unresolved? Can you locate the grounds for each alternative, including a reading you find unpersuasive? If it introduces a later event or unsupported motive, identify the addition and explain its consequence rather than silently repairing the saved output.

In the Sources and AI Register, retain the packet, task, response and examination. In the Evidence and Interpretation Map, develop the interpretation you can defend and the question it opens. In the Reflection and Learning Journal, compare that decision with your starting reasons, including a reasoned decision to keep them. One exercise can show what you made of a response; later inquiry would be needed to examine whether the reasoning carries into practice. Sections 5.6 and 6.2–6.3 develop the same source comparison and interpretive questions.

The adaptable tasks that follow let you focus on particular parts of this work. They do not require every inquiry to repeat G1’s wording or use every prompt.

**Check an excerpt before interpreting it**

**Purpose and input.** Compare a proposed extract with its permitted source passage, including the question and surrounding context. Supply the source ID and exact locator. Do not substitute a summary for the source being checked.

Adaptable instruction. “Compare the proposed extract with the supplied passage. Identify omissions or changed wording that affect its meaning. Distinguish verbatim words from paraphrase. Return the source ID, the difference, and what it leaves uncertain. Do not repair missing speech or infer context outside the supplied material.”

**Expected output and human check.** Expect a list of discrepancies or an explicit statement that none was identified. Check every comparison against the original, including speaker, negation, qualifications and what precedes the extract. A system’s failure to find an omission is not evidence that the selection is fair.

Examination and use. Place the proposed extract beside the passage and mark the differences. Keep the preparation history and any correction in the Sources and AI Register; identify affected interpretations in the Evidence and Interpretation Map. Fidelity makes interpretation possible. Deciding what the passage means requires the further work below.

In M03’s fictional answer, an extract containing only the request for a say omits the value of having someone to sit with. Restoring that opening does not weaken the request. It lets the reader see what the request could otherwise conceal about the same person’s experience.

**Pass one Present experience and imagined futures**

**Purpose and input.** Examine what people describe now and what they would welcome, retain or change. Use complete relevant passages, their questions and source locators. Include material that does not fit the proposed distinction.

Adaptable instruction. “Using only these passages, identify accounts of present experience and possible desired futures. Keep source wording separate from your interpretation. Do not turn a wish into a commitment to act. Include passages that resist the distinction and questions the material leaves open.”

**Expected output and human check.** Expect a source-linked comparison and a list of unresolved meanings. Ask whether the output has converted appreciation into satisfaction with everything, or a wish for influence into willingness to organize. Check all supplied passages, including those omitted from the output.

Examination and use. Annotate the passages using separate columns for the words, your reading, the AI proposal and the unresolved question. A participant may value continuity or want less involvement. These categories do not establish a progression toward leadership.

**Pass two Sacred language and metaphors**

**Purpose and input.** Attend to language participants themselves use for what matters deeply. Religious language may be present or absent. Supply context, not isolated words gathered from different conversations.

Adaptable instruction. “Locate sacred language or metaphors actually present in the supplied accounts. Quote only the necessary wording with its locator. Suggest possible meanings as alternatives, explain their contextual grounds, and identify what needs clarification. Do not assign a faith identity, spiritual condition or theological belief from silence or a single phrase.”

**Expected output and human check.** Expect a modest list of phrases and plausible readings, which may be empty. Check whether the output has imported your theological vocabulary into a participant’s account. Ask whose term each important word is.

Examination and use. Compare the suggested meanings with the language and circumstances supplied. M03 uses no explicitly theological term in the initial answer. The pastor can develop a hospitality interpretation while identifying it as the pastor’s argument. AI can help connect that argument to relevant theological sources in the separate task below.

**Pass three Counterexamples and alternative readings**

**Purpose and input.** Examine a provisional interpretation against relevant material that may resist it. Supply the interpretation as your proposal, not as an established result. Include the available corpus boundary and the sources being compared.

Adaptable instruction. “Compare my proposed reading with these accounts. Locate support, qualifications and contrary material, preserving each source ID and locator. Offer alternative interpretations with their grounds. Separate absence of information from contradictory evidence. State what you cannot decide from this selection.”

**Expected output and human check.** Expect a comparison that can leave the claim unresolved. Verify every supporting and contrary passage. Look for invented balance, omitted minority accounts, and false generalizations from a few selected sources. Read outside the supplied selection when the design requires it.

Examination and use. Develop the strongest alternative reading you can and compare it with the generated proposal and the sources. In the case, M02’s wish to receive dinner without another job prevents the researcher from assuming that everyone wants an organizing role. It does not establish what every attendee wants. Neither the generated proposal nor your examination supplies missing voices.

**Develop an account of learning across versions**

**Purpose and input.** Compare permitted versions of a question, assumption or claim. Supply their dates or sequence, source links, and whether each entry was contemporary or retrospective. Before asking for assistance, write your present explanation of the judgment and why it changed or remained. This gives the later comparison a human starting point (Section 3.2).

Adaptable instruction. “Compare these dated versions and their reasons. Preserve the old and new wording. Develop an account of what changed or remained and an alternative explanation. Identify the subsequent decision that would help examine each account. Distinguish revised wording, explained understanding and observed conduct.”

**Expected output and human check.** Check the comparison against the versions and their grounds in the Inquiry Register, Reflection and Learning Journal, and Ministry Experiment Journal. Identify what the tool noticed that you find useful, what you reject, and why. Explain any consequence for how you will frame, listen, or judge next time. Later practice can help examine that intention. The comparison remains an aid; neither changed wording nor a persuasive explanation establishes enduring learning.

Examination and use. Read the versions and reasons alongside the response. Follow the retained interpretation into the next decision, with its actual status recorded. J9 offers a way to examine the difference between a developed argument and enduring learning; a proposed later observation remains a plan until it occurs.

**Prepare a participant conversation**

**Purpose and input.** Turn a tentative reading into an open invitation to respond. Use a permitted summary of the interpretive issue; personal source material may not be necessary.

Adaptable instruction. “Suggest brief, non-leading questions that allow a person to correct, qualify, disagree with or decline to discuss this reading. Do not assume that the reading is true or that the person should accept a role. Include a question about what the reading has missed.”

**Expected output and human check.** Expect possible questions, not an interview script that must be followed. Read them aloud. Remove presumptions, unnecessary sensitive detail and language that makes disagreement costly. Decide whether a private route or no further approach is more appropriate.

Examination and use. Compare the suggested questions with your own wording, the actual relationship and agreed arrangements. Participant engagement needs a reason suited to the study. Birt and colleagues locate their approach within a particular epistemological and health research design; AWI does not treat it as universal certification of an interpretation (2016, accepted manuscript, 19–23).

Develop an action comparison. Supply the current question, feasible alternatives, accepted responsibilities and the observations available. Ask AI to compare what each option could illuminate, develop a plausible alternative and identify what would distinguish the possible explanations. Examine the comparison against the actual agreements and capacity. J7 demonstrates this task; a model’s preference cannot create a role someone has not accepted.

Develop a theological connection. Supply verified passages or clearly identified source summaries, the empirical account, your proposed interpretation and a consequential alternative. Ask AI to develop the reasoning between the sources, the practice and a proposed responsibility, identifying where the application extends the sources. Examine each step, return to primary passages where needed and consider whose burden the application changes. J8b demonstrates the task using summaries; its response does not certify those summaries.

Develop the report’s argument. Supply the question history, findings, methodological position, theological reasoning and proposed contribution. Ask AI to compare two organizing arguments, explain which connections are stronger and identify what each leaves unaccounted for. Examine the suggested structure against the evidence and the program’s requirements. J10 shows how this task can lead to a more precise contribution and useful defense questions.

**Write a disclosure that identifies the actual work**

This editable model should be completed from the Sources and AI Register: “For [task], I used [tool, disclosed model/version and date] with [permitted source IDs and input description]. The output proposed [kind of assistance]. I checked it against [original material], [retained, corrected or rejected] it for [reason], and used the retained portion in [record or report location]. [Explain restrictions on access to the input and output.] Responsibility for the interpretation and report remained with [researcher].”

For the book’s demonstrations, the disclosure must identify editorial work: “The Codex assistant generated a response to a fixed fictional packet on September 21, 2026. The same assistant examined it against the supplied material and used the retained reasoning in the manuscript. Existing conversation context included later case material. The exact model identifier and generation settings were not captured. No practitioner learning or field outcome was observed through this exchange.” Identify the task and its specific contribution beside this shared provenance.

For an actual AWI project, disclose the AI work actually performed and explain any redesigned or omitted task. Omitting an unsuitable operation does not remove the method’s requirement for substantive augmentation. Describe the replacement operation and its intellectual purpose, with an accurate account of any resulting limitation.

**Optional questions about context**

Use these when the inquiry needs them, without turning a conversation into a compulsory personal profile. Ask what a proposed invitation or request means in the person’s circumstances. Ask what an ordinary day makes possible or difficult. Invite the person to describe what matters to them, using their own language. Ask what they would like to preserve, change or leave alone.

These questions correspond to Chapter 4’s attention to cultural meanings, the mosaic of ordinary commitments, worldview, and present and imagined futures. They do not classify personality, ethnicity or spiritual readiness. If someone says a time does not work, record that and seek an appropriate alternative; do not supply a private explanation the person did not give.