Exercises — Chapter 21: The Road Ahead

Part A: Conceptual Questions

A1. ⭐ What is multimodal AI? Give one example of how a multimodal system could be more useful than a single-modality system.

A2. ⭐ What is the difference between an AI system that responds to prompts and an AI agent? Why does this distinction matter for safety and accountability?

A3. ⭐⭐ The chapter argues that "durable frameworks" are more valuable than knowledge of specific AI systems. In your own words, explain why. Give one example of a framework from this course that you expect to remain useful regardless of how AI evolves.

A4. ⭐⭐ Explain technological determinism. Why does the chapter argue against it?

A5. ⭐⭐ Why does the chapter present three scenarios for AI's future rather than a single prediction? What are the advantages and limitations of scenario-based thinking?

A6. ⭐⭐⭐ The chapter lists nine threshold concepts from across the textbook. Choose the one that has most changed how you think about AI. Explain what you understood before encountering it, what you understand now, and how this shift affects your evaluation of real AI systems.

A7. ⭐⭐ The chapter says, "The less visible AI becomes, the more important AI literacy becomes." Explain what this means and why you agree or disagree.

A8. ⭐⭐⭐ In your own words, summarize what "AI literacy" means. How is it different from AI expertise? Why does the distinction matter?

Part B: Applied Analysis

B1. ⭐⭐ Apply the FACTS Framework to AI agents — systems that can plan and execute multi-step tasks autonomously.

B2. ⭐⭐ Choose one of the three scenarios from section 21.3 (Augmentation Era, Muddle Through, Reckoning). For each of the four anchor examples (ContentGuard, MedAssist AI, Priya's Semester, CityScope Predict), describe what the system might look like in 2035 under that scenario.

B3. ⭐⭐⭐ The chapter identifies "a major AI failure" as a possible wild card. Design a hypothetical but plausible scenario in which an AI failure has cascading consequences across multiple domains (financial markets, healthcare, transportation, etc.). What existing safeguards might prevent or mitigate such a failure? What additional safeguards would you recommend?

B4. ⭐⭐ Review your AI Audit Report. Identify the single strongest aspect of your analysis and the single weakest. For the weakest area, what additional information or framework would you need to strengthen it?

B5. ⭐⭐⭐ The chapter argues that the internet's evolution offers lessons for AI governance. Identify three specific policy decisions (or non-decisions) about the internet that had significant consequences. For each, describe what a different choice might have produced and what the parallel AI governance decision would be.

Part C: Research Design & Critical Thinking

C1. ⭐⭐⭐ Design a survey to measure AI literacy in a population. What questions would you ask? How would you define "high," "moderate," and "low" AI literacy? What demographic variables would you control for?

C2. ⭐⭐⭐ The "muddle through" scenario suggests that AI benefits will be distributed unevenly. Design a research study that could test this hypothesis. What would you measure? How would you define "benefit" and "harm"? What populations would you compare?

C3. ⭐⭐⭐⭐ Some critics argue that "AI literacy for everyone" is impractical — that the technology is too complex for non-specialists to evaluate meaningfully. Others argue it is essential for democratic governance. Evaluate both positions. Draw on evidence from your own experience with this course: Did your ability to evaluate AI systems genuinely improve? Where do you still feel limited?

C4. ⭐⭐⭐ The chapter discusses "AI agents" as a near-term development. Identify three specific governance challenges that autonomous AI agents create that do not exist for current prompt-response AI systems. For each, propose a governance approach.

Part D: Synthesis

D1. ⭐⭐⭐ Write a 500-word "State of AI" briefing for a specific audience: choose either (a) your local city council, (b) a high school principal, or (c) the CEO of a mid-sized company. Summarize the current state of AI, the most important near-term developments, and three specific recommendations for your audience.

D2. ⭐⭐⭐⭐ Choose one recurring theme from the textbook (Tools Built by Humans, Capability vs. Understanding, Who Benefits/Who Is Harmed, Human in the Loop, AI Literacy as Civic Skill, or Durable Frameworks). Trace it across at least five chapters, showing how your understanding of the theme deepened or evolved. Conclude with a statement of why this theme matters for the future.

D3. ⭐⭐⭐ Return to the four anchor examples one final time. For each (ContentGuard, MedAssist AI, Priya's Semester, CityScope Predict), write 2–3 sentences describing the most important thing this example taught you across the entire course. What is the single lesson from each that you will carry forward?

D4. ⭐⭐⭐⭐ The chapter closes with "A Letter to the Future Reader." Write your own 300-word letter — to yourself, one year from now — about your AI literacy journey. What do you know now that you did not know when you started this course? What do you want to remember? What commitment are you making?

Part M: Mixed Practice (Final Comprehensive Review)

These questions draw from the full span of the course.

M1. ⭐⭐ (Ch.1 + Ch.21) In Chapter 1, you met the term "the AI effect." How does the AI effect relate to the public's ability to evaluate the developments described in section 21.1? Will the AI effect make it harder or easier for citizens to engage with AI policy?

M2. ⭐⭐ (Ch.3 + Ch.20) Chapter 3 introduced the threshold concept "machines learn from patterns in data, not from understanding." Chapter 20 introduced the alignment problem. How are these two ideas connected? Why does the absence of understanding make alignment harder?

M3. ⭐⭐⭐ (Ch.9 + Ch.19 + Ch.21) Synthesize the discussions of bias (Ch.9), global perspectives (Ch.19), and future scenarios (Ch.21). In 200 words, argue for or against the following claim: "AI bias is a problem that technology alone cannot solve."

M4. ⭐⭐⭐ (Ch.12 + Ch.13 + Ch.20) Drawing on privacy (Ch.12), governance (Ch.13), and safety (Ch.20), describe the ideal AI governance framework for the next decade. What principles would it be built on? What institutions would enforce it? How would it adapt to technological change?

M5. ⭐⭐ (Entire course) If you could go back and give yourself one piece of advice before starting this course, what would it be? What misconception about AI did you hold that this course corrected?