Message
All templatesAI & Technology
AI Refusal Message Clarity & Tone Evaluation
A stimulus-comparison survey for UX researchers and AI product teams to evaluate the clarity, tone, and helpfulness of AI safety and refusal messages. Produces actionable data on user preferences and improvement priorities.
No sign-up to try · Unlimited responses on the free plan
Live preview
AI Refusal Message Clarity & Tone Evaluation
Welcome, and thank you for participating in this study. This survey asks about your recent experiences with AI safety and refusal messages — the responses you see when an AI system declines a request. It should take approximately 12 minutes to complete. Your participation is entirely voluntary, and you may stop at any time. There are no right or wrong answers — we are interested only in your honest opinions. All responses are confidential, will be anonymized, and reported only in aggregate. Results will be used to improve the design of AI refusal messages. By continuing, you indicate that you understand this information and agree to participate.
Powered by QuestionPunk
Default theme. Colors and fonts are yours to change.
Sample questions
A preview of what’s in the template. Every question is editable before you launch.
23 questions · ~12 min
Q01
Multiple Choice
In the last 30 days, approximately how often have you encountered an AI safety prompt or refusal message?
- Never
- Once
- 2–3 times
- 4–6 times
- 7 or more times
Q02
Multiple Choice
In which types of AI tools or platforms have you encountered these messages? Select all that apply.
- General-purpose AI assistant (e.g., ChatGPT, Claude)
- Search engine with AI features
- Social media bot
- Customer support chatbot
- Developer tool or code assistant
- Productivity app integration
- Other (please specify)
Message
Please carefully read the following AI refusal message (Message A): I'm sorry, but I can't help with that request.
Q03
Opinion Scale
How clear was Message A?
Scale: 1 – 7
Min:Not at all clearMax:Extremely clear
Q04
Opinion Scale
How helpful is Message A in guiding you toward safe next steps?
Scale: 1 – 7
Min:Not at all helpfulMax:Extremely helpful
Q05
Multiple Choice
Which words best describe the tone of Message A? Select up to three.
- Respectful
- Helpful
- Clear
- Reassuring
- Robotic
- Vague
- Judgmental
- Condescending
Q06
Long Text
If you could edit Message A to improve it, what would you change?
Message
Please carefully read the following AI refusal message (Message B): I can't help with instructions for getting into someone else's account, because that could put their privacy and security at risk. If you're trying to regain access to your own account, I can walk you through the official account recovery steps or help you contact the service's support team.
Q07
Opinion Scale
How clear was Message B?
Scale: 1 – 7
Min:Not at all clearMax:Extremely clear
Q08
Opinion Scale
How helpful is Message B in guiding you toward safe next steps?
Scale: 1 – 7
Min:Not at all helpfulMax:Extremely helpful
Q09
Multiple Choice
Which words best describe the tone of Message B? Select up to three.
- Respectful
- Helpful
- Clear
- Reassuring
- Robotic
- Vague
- Judgmental
- Condescending
Q10
Long Text
If you could edit Message B to improve it, what would you change?
Q11
Multiple Choice
If you encountered a refusal in a real product, which message would you prefer to receive?
- Message A
- Message B
- No preference
Q12
Ranking
Rank the following qualities of a refusal or safety message from most important (1) to least important.
- Explains why it can't comply
- Offers safe alternatives
- Uses plain, simple language
- Polite and non-judgmental tone
- Brief length (concise)
- Links to trustworthy resources
Drag to rank
Q13
Multiple Choice
What types of guidance would be most helpful to you after receiving a refusal? Select all that apply.
- Step-by-step safer method
- Examples of allowed questions
- Links to trustworthy resources
- Short policy summary
- Offer to rephrase the request
- Other (please specify)
Q14
Opinion Scale
How important is it to you that an AI provides clear guidance when it cannot fulfill a request?
Scale: 1 – 7
Min:Not at all importantMax:Extremely important
Q15
Opinion Scale
Overall, how satisfied are you with the AI safety or refusal messages you have encountered in the last 30 days?
Scale: 1 – 7
Min:Very dissatisfiedMax:Very satisfied
Q16
Long Text
Briefly describe what you were trying to accomplish the last time you encountered an AI refusal message.
Q17
Long Text
Based on your responses in this survey, please share any additional thoughts or suggestions about AI safety or refusal messages.
Q18
AI Interview
We'd like to explore some of your responses in more detail. An AI moderator will ask you a few brief follow-up questions about your experiences with AI refusal messages.
Q19
Dropdown
What is your age?
- Under 18
- 18–24
- 25–34
- 35–44
- 45–54
- 55–64
- 65 or older
- Prefer not to say
Q20
Multiple Choice
What is your gender identity?
- Woman
- Man
- Non-binary
- Self-describe (please specify)
- Prefer not to say
Q21
Dropdown
In which region do you primarily live?
- Africa
- Asia
- Europe
- North America
- Oceania
- South America
- Prefer not to say
Q22
Dropdown
What is your highest level of education completed?
- Some high school or less
- High school diploma or equivalent
- Some college or Associate degree
- Bachelor's degree
- Graduate or professional degree
- Prefer not to say
Q23
Dropdown
Which of the following best describes your current employment status?
- Full-time employed
- Part-time employed
- Self-employed
- Student
- Unemployed
- Retired
- Prefer not to say
Message
Thank you for completing this survey. Your responses have been recorded and will help improve AI safety and refusal message design.
What’s included
AI follow-ups
Adaptive probes on open-ended answers that pull out detail a static form would miss.
Attention checks
Built-in safeguards against rushed answers and low-quality respondents.
AI-drafted copy
Wording, ordering, and branching written by the AI — tuned to your research goal.
Auto report
Themes, quotes, and a plain-English summary write themselves once responses come in.
Frequently asked questions
What questions are in the “AI Refusal Message Clarity & Tone Evaluation” template?
The template includes 23 ready-to-use questions, starting with: “In the last 30 days, approximately how often have you encountered an AI safety prompt or refusal message?” · “In which types of AI tools or platforms have you encountered these messages? Select all that apply.” · “How clear was Message A?”. The full set is previewed above, and every question is editable.
How long does this survey take to complete?
Respondents typically finish the 23 questions in about 12 minutes.
Can I customize this template?
Yes — every question, answer option, and the ordering is editable before you launch. You can add or remove questions, or ask the AI editor to rework the survey around your research goal.
Is this template free to use?
Yes. Open it in the editor and start customizing right away — no account required to try it, and the free plan covers launching your survey.
Ready to launch?
Open this template in the editor. Every part is yours to change before the first respondent sees it.
Related templates
More studies on similar topics.
