Quiz: Ethics and Governance
Test your knowledge about AI ethics, copyright, and regulation.
30 questions - Click on each question to reveal the correct answer.
Questions 1-30
Question 1: What is AI bias?
A) A type of AI B) Systematic errors causing unfair outcomes C) A training technique D) A model size
Answer: B - Bias refers to systematic errors that produce unfair outcomes for certain groups.
Question 2: Where does training data bias come from?
A) Random noise B) Historical patterns and underrepresentation C) Computer errors D) User inputs only
Answer: B - Training data bias comes from historical patterns, stereotypes, and underrepresentation.
Question 3: What is the EU AI Act?
A) A movie B) Comprehensive AI regulation C) An AI model D) A training dataset
Answer: B - The EU AI Act is the most comprehensive AI regulation globally.
Question 4: What risk level prohibits certain AI uses in the EU?
A) Minimal B) Limited C) High D) Unacceptable
Answer: D - Unacceptable risk AI applications (like social scoring) are prohibited.
Question 5: Who owns AI-generated content in the US?
A) The AI B) Generally the user C) The government D) No one ever
Answer: B - Generally, the user who creates AI-generated content owns the output.
Question 6: What does the US Copyright Office say about AI-only works?
A) Fully copyrightable B) Not copyrightable without human authorship C) Always public domain D) Owned by AI companies
Answer: B - Works created solely by AI without human authorship are not copyrightable.
Question 7: What is demographic parity?
A) Population counting B) Equal outcome rates across groups C) Age distribution D) Geographic location
Answer: B - A fairness metric where outcomes are distributed equally across demographic groups.
Question 8: What is AI alignment?
A) Positioning models B) Ensuring AI acts according to human values C) Aligning text D) Hardware setup
Answer: B - Ensuring AI systems behave in accordance with human values and intentions.
Question 9: What should organizations document about AI?
A) Nothing B) Systems, data, decisions, processes C) Only errors D) Only successes
Answer: B - Document AI systems, training data, decision processes, and impacts.
Question 10: What is required for high-risk AI under EU rules?
A) Nothing special B) Strict requirements and human oversight C) Only disclosure D) Free access
Answer: B - High-risk AI requires strict requirements including human oversight.
Question 11: What is selection bias?
A) Choosing AI models B) Training data not representing full population C) User selections D) Feature selection
Answer: B - When training data doesn't adequately represent all groups.
Question 12: What major lawsuits involve AI training data?
A) None exist B) NYT, Getty, Authors vs AI companies C) Only personal D) Only hardware
Answer: B - Multiple lawsuits challenge AI training on copyrighted content.
Question 13: What is NIST AI RMF?
A) A company B) AI Risk Management Framework C) A neural network D) A regulation
Answer: B - NIST's voluntary AI Risk Management Framework.
Question 14: When should AI use be disclosed?
A) Never B) Increasingly in many contexts C) Only errors D) Only if asked
Answer: B - Disclosure is increasingly expected or required in many contexts.
Question 15: What helps establish copyright for AI-assisted work?
A) More AI use B) Substantial human creative input C) Hiding AI use D) Faster generation
Answer: B - Human creative contribution helps establish copyright claims.
Question 16: What approach does the EU AI Act use?
A) Size-based B) Risk-based C) Cost-based D) Age-based
Answer: B - The EU AI Act uses a risk-based approach to regulation.
Question 17: What is hallucination in AI ethics?
A) Visual effects B) Generating plausible but false information C) Dreams D) Errors only
Answer: B - AI generating confident but incorrect or fabricated content.
Question 18: How should organizations prepare for AI regulation?
A) Ignore it B) Inventory, assess, document, control C) Wait and see D) Stop using AI
Answer: B - Inventory AI systems, assess risks, document, and implement controls.
Question 19: What does transparency in AI require?
A) Open source code B) Disclosing AI use and explaining decisions C) Free access D) Sharing profits
Answer: B - Being open about AI use and how AI-driven decisions are made.
Question 20: What is algorithmic bias?
A) Algorithm speed B) Unfair patterns from model design C) Code bugs D) Data storage
Answer: B - Bias introduced through model design and optimization choices.
Question 21: What does human oversight mean for AI?
A) Humans building AI B) Humans maintaining decision authority over AI C) Human users D) Human errors
Answer: B - Maintaining human control and review over AI decisions.
Question 22: What region has required government approval for AI?
A) US B) China C) Canada D) Australia
Answer: B - China requires government approval for certain AI services.
Question 23: What is an AI audit?
A) Financial review B) Systematic evaluation of AI for bias/issues C) Sound check D) User survey
Answer: B - Systematic evaluation of AI systems for bias, fairness, and compliance.
Question 24: How should disclosure be phrased?
A) Hide AI use B) "Created with AI assistance" or similar C) No disclosure D) Claim human-only
Answer: B - Clear, honest disclosure like "Created with AI assistance."
Question 25: What helps detect AI bias?
A) Ignoring data B) Testing across demographic groups C) Using more AI D) Faster processing
Answer: B - Testing and comparing outcomes across different demographic groups.
Question 26: What is FTC's role in AI?
A) Building AI B) Enforcing against deceptive practices C) Training AI D) Selling AI
Answer: B - The FTC enforces against deceptive AI practices in the US.
Question 27: What does GDPR affect in AI?
A) Only websites B) Training data and personal data use C) Only images D) Only text
Answer: B - GDPR affects how personal data can be used in AI training.
Question 28: What is red team testing?
A) Color coding B) Adversarial testing to find problems C) Team building D) Sports
Answer: B - Adversarial testing where teams try to find flaws and biases.
Question 29: What should diverse development teams help with?
A) Nothing B) Identifying and reducing bias C) Faster coding D) Lower costs
Answer: B - Diverse teams help identify potential biases others might miss.
Question 30: What is the key to responsible AI?
A) Fastest AI B) Ongoing attention to ethics, transparency, and fairness C) Biggest AI D) Cheapest AI
Answer: B - Continuous commitment to ethical use, transparency, and fairness.
Results
| Score | Level |
|---|---|
| 27-30 | Excellent - Strong understanding of AI ethics |
| 21-26 | Good - Some review may be helpful |
| 15-20 | Average - Review the chapters again |
| Below 15 | Needs comprehensive review |