Module Check: Governing AI
You have covered ten lessons on one of the most consequential questions of our time: how do human societies govern a technology that is advancing faster than institutions can adapt? Before you move on, let us consolidate what you know. This module check opens with a full-module vocabulary review, then tests your analytical understanding with comprehensive quiz questions drawn from across all lessons. It closes with a synthesis challenge that asks you to apply everything together.
Flashcards — click each card to reveal the answer
Module Quiz
A country's parliament is debating whether to regulate an AI system used by its courts to recommend sentencing lengths for convicted offenders. One faction argues the system is technically accurate and should be deployed without restriction. The opposing faction argues that accuracy alone is insufficient for deployment in this context. Which of the following best captures what the second faction is arguing?
The EU AI Act's provisions apply to a US-based AI company that deploys its system to European users for use in CV screening. The company argues it is not subject to EU law because it has no offices in the EU. Is the company correct?
An AI auditor tests a credit-scoring model and finds no statistically significant disparate impact across racial groups on their representative test dataset. The auditor certifies the model as compliant with fairness requirements. Six months later, the model is found to produce significant disparate impact for a specific immigrant subgroup not adequately represented in the test dataset. Which limitation of auditing does this illustrate?
The United States and China are both developing frontier AI systems with significant military applications. Both have signed the Bletchley Declaration committing to information-sharing about catastrophic AI risks. A researcher proposes that both countries should share detailed technical information about their most capable military AI systems as a confidence-building measure. What structural obstacle does this proposal face?
A government agency is deciding whether to deploy a predictive policing AI that increases overall crime detection rates by 15% but generates false-arrest recommendations for Black residents at three times the rate of white residents. A consequentialist analysis supports deployment on aggregate-benefit grounds. Which ethical framework provides the strongest challenge to this conclusion, and what is its core argument?
A high school student wants to contribute to AI governance and is deciding between two paths: (A) becoming a machine learning researcher working on technical alignment, or (B) studying public policy with a focus on technology regulation. Which of the following most accurately reflects the state of need in the AI safety and governance field?
Synthesis: The Governance Memo
- You are a senior policy advisor to a newly elected government. The Prime Minister has asked for a two-page briefing memo answering: 'What are the three most important actions our government should take in the next 18 months to govern AI responsibly?'
- Your memo must draw on specific material from across this module. It must:
- 1. Identify three distinct governance actions (they may be legislative, regulatory, international, standards-based, or research-support actions — but they should not all be the same type).
- 2. For each action: state what it is, explain why it is among the three most important (referencing specific risks or failures from the module), identify the key actor responsible for executing it, and acknowledge the most serious objection to it.
- 3. Acknowledge one area where you do not have a good answer yet — a genuine open question in AI governance your government faces.
- 4. Conclude with one sentence explaining what role ordinary citizens — not just policymakers and engineers — should play in AI governance.
- Your memo is graded on: specificity (no vague aspirations), honest acknowledgment of trade-offs, clear connection to module concepts, and persuasiveness.
- Extension: Share memos with the class. Vote on the most persuasive recommendation. Discuss: where did memos agree? Where did they diverge, and what does that reveal about genuine uncertainty in the field?