AI Fluency (Advanced User) Assessment: Sample Questions | Bryq
Ethical & Responsible Use
AI Task Strategy
Prompting & Interaction Quality
Critical Evaluation & Validation
Workflow Integration & Output Quality
These are five sample questions from the Advanced User of Bryq’s AI Fluency (MCQ) assessment, a test for hiring employees and team leads who design and guide AI-assisted work. The full assessment covers five skills: AI task strategy, prompting and interaction quality, critical evaluation and validation, ethical and responsible use, and workflow integration and output quality. At this level, questions focus on designing AI-assisted workflows, evaluating outputs systematically, producing professional work with AI support, and coaching others on effective AI use. Each question describes a complex workplace situation and asks which response fits best, with four options to choose from.
Ethical & Responsible Use
Advanced
Question
Your team wants to use an AI tool to rank job applicants’ CVs and only review the top 20%. What should happen before it’s rolled out?
Select your answer
Test the tool’s rankings for bias across candidate groups, keep people accountable for decisions and check legal requirements.
Use the tool, but have a recruiter review a random sample of rejected CVs each month.
Use the tool as planned, since it applies the same criteria to every applicant consistently.
Use the tool, but remove names from CVs first so its rankings can’t be biased.
WHY IS THIS THE CORRECT ANSWER
AI used in hiring can reproduce historical bias through indirect signals such as schools, career gaps or locations, even when names are removed, and many jurisdictions regulate automated hiring decisions. Testing for bias before launch, keeping people accountable for decisions and checking legal obligations addresses the risk before any candidates are affected.
AI Task Strategy
Advanced
Question
You’re asked to produce a pricing recommendation for a new product, using AI where it helps. How should the work be split between you and the AI?
Select your answer
Use AI to gather competitor pricing and draft scenarios, while you set the assumptions and make the final recommendation.
Have the AI produce the full recommendation, then present it after a careful proofread for clarity and tone.
Use AI only to format the final slides, and keep all research and analysis manual to avoid errors.
Use AI for research and scenario modeling, and let it choose between the scenarios using criteria you define.
WHY IS THIS THE CORRECT ANSWER
AI can speed up research, structuring and scenario drafting, but a pricing decision depends on assumptions and trade-offs that a person must own. Keeping the assumptions and the final judgment with you uses AI where it adds speed, while keeping accountability for the decision.
Prompting & Interaction Quality
Advanced
Question
You use an AI assistant to turn a raw data export into a weekly sales summary, but the format of its output changes every week. What is the most effective fix?
Select your answer
Give the AI a fixed output template with defined sections, an example of a good summary and explicit rules for edge cases.
Tell the AI each week to “use the same format as last time” before pasting in the new export.
Ask for a summary each week, then manually reformat the output to match last week’s version.
Try a different AI tool each week until one produces a format the team is happy with.
WHY IS THIS THE CORRECT ANSWER
AI output varies when instructions leave room for interpretation, and most assistants don’t reliably remember previous sessions. A fixed template, an example and explicit rules remove that ambiguity, so each week’s summary comes out consistent without manual rework.
Critical Evaluation & Validation
Advanced
Question
For three months, an AI tool that screens supplier invoices for errors has flagged almost nothing. A manual audit then finds several problems the tool missed. What does this most likely indicate?
Select your answer
The tool may be missing certain types of error, so test it against known errors before trusting its results again.
The invoices are mostly accurate, and the audit simply found rare exceptions the tool understandably missed.
The tool needs a software update, after which its results can be trusted without further testing.
Manual audits should run more often, while the tool continues to handle day-to-day checks as before.
WHY IS THIS THE CORRECT ANSWER
A tool that rarely flags anything can look reliable while quietly missing whole categories of errors. Testing it against invoices with known errors measures what it actually catches, showing where its results can be trusted and where human checks are still needed.
Workflow Integration & Output Quality
Advanced
Question
As your team’s AI lead, you notice that colleagues each write their own prompts for the same weekly client report, producing inconsistent quality and occasional factual errors. What is the best way to address this?
Select your answer
Build a shared prompt template with required inputs and a review checklist, test it with the team and refine it.
Write one strong prompt yourself, and ask everyone to copy and paste it exactly each week.
Send the team a guide to general prompting best practices, and let each person improve their own approach.
Take over the weekly client report yourself, since you’re the most experienced AI user on the team.
Next
WHY IS THIS THE CORRECT ANSWER
A shared template captures what makes the report work, so quality no longer depends on each person’s prompting skill. Building in a review checklist targets the factual errors, and testing the template with the team makes sure it fits how they actually work. A fixed prompt with no review step, or generic advice, leaves the underlying problems in place.
