Short answer
Responsible AI Quiz checks whether the learner can transfer ideas from the module into new situations. It is not a memory race. Each question asks for the safest, most evidence-based or most mathematically justified action. The four options are shuffled every time the quiz begins, so the correct answer does not remain in one screen position.
Why this assessment matters
This responsible AI assessment checks whether the learner can transfer evidence, human oversight, privacy and risk into unfamiliar decisions rather than remember where an answer appeared. A useful score points to a specific review action: revisit the relevant lesson, reproduce the evidence artefact and explain the boundary in a new scenario.
Learning objectives
- Apply module concepts in unfamiliar scenarios.
- Distinguish evidence-based actions from confident guesses.
- Use explanations to identify the reason behind each answer.
- Create a follow-up practice task from an incorrect or uncertain response.
Interactive quiz
Read each responsible AI scenario by locating the claim, the affected people, the evidence gap and the human decision point. Choose the option that preserves privacy, verification and accountability. The options are reshuffled on every start and restart.
Common quiz mistakes
- Choosing an option because it repeats a word from the question.
- Treating the longest answer as automatically correct.
- Ignoring the safety, evidence or unit condition in the scenario.
- Looking only at the score and skipping the explanations.
Review strategy and summary
Before starting the responsible AI quiz, build a one-page map of the module: one core idea, one observable example, one test method and one limitation for each lesson. After the quiz, classify uncertain answers as a concept gap, an evidence gap, a safety or accessibility gap, or rushed reading; use that label to choose the next practice task.
Review questions
- Which action best applies “retrieval versus generation” in the context of Generative AI and Search Engines: What Is the Difference??
- Which action best applies “missing or weak grounding” in the context of Why AI Can Make Things Up?
- Which action best applies “constraints and exclusions” in the context of Define the Problem Before Writing the Prompt?
- Which action best applies “security and licence review” in the context of Verifying AI-Assisted Code?
- Which action best applies “data classification” in the context of Personal and Confidential Data in AI Tools?
- Which action best applies “original source search” in the context of Deepfakes and Content Provenance?
- Which action best applies “reversibility” in the context of Human Oversight and Levels of Risk?
- Which action best applies “governance and monitoring” in the context of A Responsible AI Project Canvas?
- Which action best applies “claim inventory” in the context of Project: Auditing an AI Answer with Evidence?
- Which action best applies “false positive and false negative” in the context of Project: Investigating Classification Errors?
- Which action best applies “source traceability” in the context of Generative AI and Search Engines: What Is the Difference??
- Which action best applies “verification and uncertainty” in the context of Why AI Can Make Things Up?
Answers with explanations
- Which action best applies “retrieval versus generation” in the context of Generative AI and Search Engines: What Is the Difference??
The correct choice uses retrieval versus generation as a decision rule and keeps the evidence trail visible.
- Which action best applies “missing or weak grounding” in the context of Why AI Can Make Things Up?
The correct choice uses missing or weak grounding as a decision rule and keeps the evidence trail visible.
- Which action best applies “constraints and exclusions” in the context of Define the Problem Before Writing the Prompt?
The correct choice uses constraints and exclusions as a decision rule and keeps the evidence trail visible.
- Which action best applies “security and licence review” in the context of Verifying AI-Assisted Code?
The correct choice uses security and licence review as a decision rule and keeps the evidence trail visible.
- Which action best applies “data classification” in the context of Personal and Confidential Data in AI Tools?
The correct choice uses data classification as a decision rule and keeps the evidence trail visible.
- Which action best applies “original source search” in the context of Deepfakes and Content Provenance?
The correct choice uses original source search as a decision rule and keeps the evidence trail visible.
- Which action best applies “reversibility” in the context of Human Oversight and Levels of Risk?
The correct choice uses reversibility as a decision rule and keeps the evidence trail visible.
- Which action best applies “governance and monitoring” in the context of A Responsible AI Project Canvas?
The correct choice uses governance and monitoring as a decision rule and keeps the evidence trail visible.
- Which action best applies “claim inventory” in the context of Project: Auditing an AI Answer with Evidence?
The correct choice uses claim inventory as a decision rule and keeps the evidence trail visible.
- Which action best applies “false positive and false negative” in the context of Project: Investigating Classification Errors?
The correct choice uses false positive and false negative as a decision rule and keeps the evidence trail visible.
- Which action best applies “source traceability” in the context of Generative AI and Search Engines: What Is the Difference??
The correct choice uses source traceability as a decision rule and keeps the evidence trail visible.
- Which action best applies “verification and uncertainty” in the context of Why AI Can Make Things Up?
The correct choice uses verification and uncertainty as a decision rule and keeps the evidence trail visible.
Privacy and data note
The responsible AI quiz runs entirely in the browser and does not send answers or scores to a server. Use fictional or redacted data and explicit human review; keep names, passwords, precise locations, private messages and unpublished project evidence out of notes and screenshots.
Review focus 1: Generative AI and Search Engines: What Is the Difference?
Search engines retrieve and rank indexed sources, while generative AI produces new responses from learned patterns and may not preserve a reliable source trail. A strong review connects retrieval versus generation with ranking and synthesis, then uses source traceability to create evidence and task-appropriate tool choice to state a limit. Practise by considering this situation: A learner needs the current rules of a competition and receives a fluent AI answer without a date or official source. Your review artefact should compare a search workflow with an AI-assisted workflow and document which claims can be traced to primary evidence.
Review focus 2: Why AI Can Make Things Up
Generative AI can produce plausible but unsupported details because it predicts language rather than checking every statement against reality. A strong review connects probabilistic generation with missing or weak grounding, then uses fabricated citations to create evidence and verification and uncertainty to state a limit. Practise by considering this situation: An AI response gives a convincing library name and documentation link that do not exist. Your review artefact should audit a synthetic answer claim by claim, classify support levels and rewrite it with verified evidence and uncertainty labels.
Review focus 3: Define the Problem Before Writing the Prompt
A strong prompt cannot repair a poorly defined problem; purpose, audience, constraints, evidence and success criteria must come first. A strong review connects problem boundary with desired outcome, then uses constraints and exclusions to create evidence and evaluation criteria to state a limit. Practise by considering this situation: A student asks an AI system to “make the best robot project” without defining age, materials, time or safety limits. Your review artefact should write a problem brief first, then produce and compare two prompts derived from the same brief.
Review focus 4: Verifying AI-Assisted Code
AI-assisted code should be understood, tested and reviewed before use because it may contain logical errors, insecure patterns or invented dependencies. A strong review connects line-by-line understanding with test cases and edge cases, then uses dependency verification to create evidence and security and licence review to state a limit. Practise by considering this situation: Generated code appears to work for one input but fails silently when a sensor returns a missing or extreme value. Your review artefact should create a verification checklist, unit-style tests and a corrected version with documented changes.
Review focus 5: Personal and Confidential Data in AI Tools
Personal, confidential or third-party information should not be placed into AI tools without a clear need, permission and understanding of the service’s data practices. A strong review connects data classification with purpose limitation, then uses consent and third-party rights to create evidence and redaction and safer substitutes to state a limit. Practise by considering this situation: A project report contains names, school details, private messages and precise location data before being pasted into a chatbot. Your review artefact should classify each data field, remove unnecessary identifiers and produce a privacy-safe test version.
Review focus 6: Deepfakes and Content Provenance
Deepfakes and synthetic media require source, context and provenance checks because visual or audio realism is not proof of authenticity. A strong review connects content clues and limits with original source search, then uses provenance credentials to create evidence and independent corroboration to state a limit. Practise by considering this situation: A dramatic video appears to show a known person making an urgent announcement, but only reposted clips are available. Your review artefact should build a verification record that separates visual observations, source history, provenance signals and unresolved questions.
Review focus 7: Human Oversight and Levels of Risk
Human oversight should become stronger as possible harm, uncertainty, scale and difficulty of correction increase. A strong review connects risk severity with likelihood and exposure, then uses reversibility to create evidence and meaningful human control to state a limit. Practise by considering this situation: An AI system suggests a spelling correction in one case and recommends access to a school opportunity in another. Your review artefact should classify both uses by risk and design a human review, appeal and stop mechanism for the higher-risk case.
Review focus 8: A Responsible AI Project Canvas
A responsible AI project canvas makes purpose, affected people, data, failure modes, safeguards, evaluation and human decisions visible before building. A strong review connects purpose and necessity with people and impacts, then uses data and model limits to create evidence and governance and monitoring to state a limit. Practise by considering this situation: A team proposes an AI classifier because it sounds impressive, but has not shown why simpler rules are insufficient. Your review artefact should complete a project canvas, compare an AI and non-AI option and define evidence required before deployment.
Turn the score into the next learning decision
When the quiz ends, the result is stored only in this browser. It is not sent to a server, no account is created and nothing is synchronised across devices.
A score of 90 or above suggests a 30-day return, 70–89 a seven-day return, and a lower score a next-day return. Missed questions can be retried in a separate session.
The progress centre shows best score, latest attempt, upcoming review and difficult questions. Local history can be cleared for one quiz or for all quizzes.
Sources and verification note
The official or primary references listed below provide the technical and educational foundation for “Responsible AI Quiz”. These links support the concepts; they do not prove that a proposed project has been physically completed. Dates, software behaviour and policy details should be rechecked before future publication updates.
- UNESCO — AI Competency Framework for Students
- NIST — Artificial Intelligence Risk Management Framework 1.0
- NIST — Generative AI Profile for the AI Risk Management Framework
Next step
For “Responsible AI Quiz”, return to the module page, complete the evidence artefact for this lesson and continue to the next item in sequence. For “Responsible AI Quiz”, a project should be presented as completed personal work only after real testing evidence and publication approval exist.