Create an FAQ from a Notice with AI: Leave Unsupported Answers as Needs Confirmation
This exercise turns a short notice into six FAQ entries. Questions supported by the source receive source IDs, while topics such as fees, refunds, and waitlists that do not appear in the notice are left as needs confirmation instead of being answered from assumptions.
Content checked 2026.09.23Copyable prompts
Show contents
Who this is forBeginners who want to organize common questions from an event, class, or meeting notice but are concerned that AI may invent answers beyond the source
What you need
You only need to be able to separate a notice into individual statements.
Add source IDs such as S1 and S2 before each statement.
No real personal data or actual organization information is required.
01Turn the notice into a fact list with source IDs
Before creating an FAQ, separate the facts that can actually be confirmed from the notice. The event in this exercise is entirely fictional, and only S1–S4 should be treated as official input. In particular, there are no statements about participation fees, refunds, or a waitlist. Explicitly marking these gaps helps prevent AI from filling them with common event practices.
Sample input
S1 Event date and time: 11-20 at 16:00
S2 Participation: advance registration required, capacity 16 people
S3 What to bring: writing utensils
S4 Online participation: not provided
Items not included in the source:
- No fee information
- No refund information
- No waitlist information
source ID
Source fact
What the FAQ can answer
S1
11-20 at 16:00
Event date and start time
S2
Advance registration required, capacity 16 people
Whether registration is required and capacity
S3
Writing utensils
What to bring
S4
Online participation not provided
Whether online participation is available
A source ID is not decoration. It is a verification marker. Someone reading the FAQ should be able to see S2 and immediately locate the original S2 statement.
02Design questions that include both answerable and unresolved topics
An FAQ does not need a definite answer for every question. You can include questions that readers are likely to ask and mark them as needs confirmation when the source does not contain the answer. In this exercise, you will create exactly six questions, with fees and refunds combined into one question.
Prompt
Create 6 FAQ entries from the notice below.
Rules:
- Use only information that actually appears in S1–S4.
- If an answer has supporting evidence, show the source ID at the end.
- If the source does not contain the answer, do not guess. Mark it as needs confirmation.
- Preserve the fact that fee, refund, and waitlist information is not provided in the source.
- Do not change online participation not provided into online participation being available.
- Do not change the date, time, or capacity number.
Use these 6 FAQ questions:
1. When is the event held?
2. Is advance registration required, and what is the capacity?
3. What should I bring?
4. Can I participate online?
5. What are the participation fee and refund policy?
6. If capacity is full, can I join a waitlist?
Because the questions themselves are fixed in this prompt, AI is less likely to add extra questions or omit important facts. Marking an answer as unknown is not a failure. It is a valid result when the source does not support an answer.
03Compare the FAQ example against the notice
The table below is not actual output from a particular AI service and does not measure AI performance. It is an editorial example created and reviewed with AI assistance to demonstrate the relationship between the input and answers. In real use, each answer should be checked directly against the referenced source ID.
Example result
Q1. When is the event held?
A. It is held on 11-20 at 16:00. Evidence: S1
Q2. Is advance registration required, and what is the capacity?
A. Advance registration is required, and the capacity is 16 people. Evidence: S2
Q3. What should I bring?
A. Please bring writing utensils. Evidence: S3
Q4. Can I participate online?
A. Online participation is not provided. Evidence: S4
Q5. What are the participation fee and refund policy?
A. The source does not provide fee or refund information. Needs confirmation.
Q6. If capacity is full, can I join a waitlist?
A. The source does not provide waitlist or waitlist registration information. Needs confirmation.
Question
Answer summary
Evidence
Status
When is the event held?
11-20 at 16:00
S1
Supported
Is registration required, and what is the capacity?
Advance registration required, capacity 16 people
S2
Supported
What should I bring?
Writing utensils
S3
Supported
Can I participate online?
Not provided
S4
Supported
What are the fee and refund policy?
No information in source
None
Needs confirmation
Can I join a waitlist?
No information in source
None
Needs confirmation
04Filter out plausible but unsupported FAQ answers
The main risk in automatic FAQ creation is turning common real-world practices into source-backed facts. Free admission, no-refund policies, and waitlist registration may be common in some situations, but none of them appear in this fictional notice. Plausible does not mean supported.
Flawed result
Q. What are the participation fee and refund policy?
A. Participation is free, so there is no separate refund process.
Q. If capacity is full, can I join a waitlist?
A. Yes. Once the 16-person capacity is reached, applicants are automatically placed on the waitlist.
Invented detail
Source status
Correct treatment
Participation is free
No fee information
Needs confirmation
No separate refund process
No refund information
Needs confirmation
Automatic waitlist registration
No waitlist information
Needs confirmation
When revising, removing unsupported claims is more important than making the answer sound polished. Even if you want to add contact information after needs confirmation, do not invent a phone number or email address if the source does not contain one.
05Re-request correction only for answers without source support
When an unsupported FAQ answer appears, specify exactly which statement goes beyond the source. Do not ask AI to fill the gap with new information. Instead, ask it to clearly mark the information as absent from the source. Any answer without a source ID should be treated as something to review again.
Prompt
Revise only Q5 and Q6 from the previous FAQ.
The claims in Q5 that participation is free and that there is no refund process do not appear in the source. Change both fee and refund information to needs confirmation.
The automatic waitlist registration in Q6 also does not appear in the source. Change waitlist availability to needs confirmation.
Do not invent contact details, policies, or exceptions that are not in S1–S4.
Do not change the date, time, 16-person capacity, writing utensils, or online participation not provided in Q1–Q4.
This kind of re-request locks the response back to the source. Even after correction, check that Q1–Q4 were not unintentionally changed.
06Check the FAQ by source ID before publishing
An FAQ should be easy to read, but it should not contain more facts than the original notice. You can rephrase questions naturally, but changing dates, capacity, negative conditions, or unresolved items turns it into a different notice. Check each item below.
Checklist
[ ] Are there exactly 6 FAQ entries?
[ ] Does 11-20 at 16:00 match S1?
[ ] Do advance registration required and capacity 16 people match S2?
[ ] Is the required item still writing utensils?
[ ] Is the negative wording online participation not provided preserved?
[ ] Does the FAQ avoid saying free or paid when no fee information exists?
[ ] Does the FAQ avoid inventing a refund policy when none is provided?
[ ] Does the FAQ avoid guessing whether a waitlist is available?
[ ] Are unsupported answers marked needs confirmation?
When applying this method to a real notice, unsupported questions can also be collected in a separate list for a person in charge to verify before updating the source notice itself. This tutorial does not determine real organization policies or legal obligations. It only practices preserving facts within the fictional notice provided.
What to check yourself
Editorial example in which six FAQ entries were checked by a human against the fictional S1–S4 event notice and the explicitly missing fee, refund, and waitlist information
Check that there are exactly 6 FAQ entries
Check that the date, time, capacity, what-to-bring item, and online participation condition from S1–S4 are preserved
Check that no fee, refund, or waitlist information was invented
Check that unsupported questions are marked needs confirmation
Verification limits
The result is an educational editorial example created and reviewed with AI assistance. It is not a result of executing an actual AI product, measuring response accuracy, or verifying any specific service's FAQ-generation feature.
This exercise uses AI to compare V1 and V2 of the same event notice and classify five comparison items as changed, added, deleted, or unchanged. You will learn how to record only differences that actually appear in the documents without assuming why a deleted sentence disappeared.
This exercise classifies six short free-text comments into predefined themes and allows multiple labels when one response contains more than one topic. It also explains why the total number of labels can exceed the number of respondents and how to avoid unsupported conclusions about satisfaction or emotional intensity.
This exercise rewrites four fictional borrowing rules in clearer language while preserving the registered-member requirement, the 7-day deadline, the prohibition on extending reserved materials, and the rule that non-reserved materials may be submitted for one extension request without guaranteeing approval.
This exercise shows how to break a vague research goal into factual questions, experience questions, and decision questions, then plan what to verify and where to look. Instead of asking AI to invent search results, you first organize the sources to check and the items that are still unknown.
This exercise shows how to design a 15-minute user interview for improving a fictional application form. You will revise leading questions into neutral questions about recent real behavior and learn not to treat predicted responses as evidence before any interviews have been conducted.
This exercise shows how to define constraints such as participant count, time, printed materials, purchasing limits, and available equipment before asking AI for 4 distinct alternatives. Instead of deciding which idea is best, you check whether each option satisfies the constraints and learn how to revise the request when the constraints change.
This exercise shows how to prepare 6 expected audience questions with source evidence, short answers, and a verification plan when the answer is unknown. It does not cover creating a presentation outline or AI voice evaluation; the focus is on answering questions without overstating limited evidence.