Quiz — Chapter 1: What Is Artificial Intelligence?

Instructions: Complete all sections. For multiple choice, select the single best answer. For true/false, provide a 1–2 sentence justification. For short answer and applied scenarios, aim for clear, concise responses.

Scoring Guide: - Section 1 (Multiple Choice): 2 points each (20 points total) - Section 2 (True/False with Justification): 3 points each (15 points total) - Section 3 (Short Answer): 5 points each (20 points total) - Section 4 (Applied Scenario): 10 points each (10 points total) - Total: 65 points

Section 1: Multiple Choice (2 points each)

1. Which of the following best describes artificial intelligence as defined in this chapter?

a) A computer system that thinks and feels like a human b) Computer systems that perform tasks typically associated with human intelligence c) Any computer program that runs automatically d) A single technology that powers all modern software

Answer **b)** The chapter defines AI as "the design of computer systems that perform tasks typically associated with human intelligence — such as recognizing patterns, making predictions, understanding language, or making decisions." The definition emphasizes task performance, not human-like thought or consciousness.

2. What is the primary difference between narrow AI and general AI?

a) Narrow AI is less expensive to build than general AI b) Narrow AI is designed for specific tasks while general AI would have human-like cognitive flexibility across domains c) Narrow AI uses older technology while general AI uses newer technology d) Narrow AI only works on text while general AI works on images and video

Answer **b)** Narrow AI excels at specific tasks within defined boundaries, while general AI (hypothetical) would demonstrate human-like adaptability and reasoning across all domains without task-specific training.

3. Which of the following is an example of the AI effect?

a) An AI system performing better than expected b) People saying that chess-playing computers are "not really AI" after Deep Blue defeated Kasparov c) AI systems becoming more accurate over time d) Companies using the term "AI" in their marketing

Answer **b)** The AI effect is the tendency to redefine intelligence to exclude whatever machines can already do. Once Deep Blue won, people dismissed chess as "just computation" rather than acknowledging it as AI.

4. In the FACTS Framework, what does the "T" stand for?

a) Technology b) Timeline c) Training d) Testing

Answer **c)** The T stands for Training — "What data was it trained on, and who curated that data?" This question helps reveal potential sources of bias and limitation in AI systems.

5. Which statement about current AI systems is accurate as of 2026?

a) General AI exists but is only available to governments and large corporations b) All AI systems in use today are examples of narrow AI c) Most AI systems can transfer knowledge freely between domains d) AI systems now understand language in the same way humans do

Answer **b)** The chapter states explicitly: "as of 2026, general AI does not exist. Every AI system you interact with today — every single one — is narrow AI."

6. What concern does the chapter raise about MedAssist AI's real-world deployment?

a) It was too expensive for the hospital to maintain b) It performed less accurately for patients with darker skin tones on dermatological images c) It was unable to read any medical images d) Doctors refused to use the system

Answer **b)** The chapter notes that MedAssist's accuracy was "significantly lower for patients with darker skin tones when analyzing dermatological images," illustrating how AI systems can perform unevenly across demographic groups.

7. Why does the chapter describe AI literacy as a "civic skill"?

a) Because all citizens are required to learn about AI by law b) Because AI systems increasingly shape decisions in healthcare, policing, hiring, and other areas affecting public life c) Because AI was invented by the government d) Because you need AI literacy to vote in elections

Answer **b)** The chapter argues that as AI systems shape hiring, healthcare, education, policing, and information access, understanding them is "a prerequisite for meaningful participation in democratic society."

8. What is the main problem with the CityScope Predict system described in the chapter?

a) It is too expensive for the city of Millhaven b) It could create a feedback loop by using historically biased crime data to direct future policing c) It replaces all police officers with robots d) It only works in large cities

Answer **b)** The chapter explains that historical crime data reflects decades of racially biased policing. If certain neighborhoods were over-policed, the data shows more crime there, and the algorithm recommends more policing — creating a self-reinforcing feedback loop.

9. Which of the following best represents the chapter's threshold concept?

a) AI will eventually surpass human intelligence in all domains b) AI is a spectrum of techniques, not a single technology c) All AI systems use neural networks d) AI is too complex for non-experts to understand

Answer **b)** The threshold concept is that there is no single "AI" — there are many different techniques (rule-based systems, statistical models, neural networks, etc.) each suited to different problems. Understanding this changes how you evaluate all AI claims.

10. A large language model (LLM) like ChatGPT is best described as:

a) A general AI system because it can handle many different types of questions b) A narrow AI system that is very versatile within a single modality (language) c) Neither narrow nor general AI d) A human-level intelligence that understands what it writes

Answer **b)** The chapter notes that while LLMs can handle a wide range of language tasks, they are "still fundamentally doing one thing — predicting the next token in a sequence." Their versatility is "breadth within a single modality, not the cross-domain flexibility that would characterize true general intelligence."

Section 2: True/False with Justification (3 points each)

For each statement, write "True" or "False" and provide a 1–2 sentence justification citing evidence from the chapter.

11. A chess-playing computer that can defeat any human grandmaster is an example of general AI.

Answer **False.** A chess engine is narrow AI — it is designed to perform one specific task (playing chess) and cannot do anything outside that domain. The chapter notes that such a system "cannot play checkers unless separately programmed to do so."

12. The FACTS Framework requires technical expertise in computer science to use effectively.

Answer **False.** The chapter explicitly states: "You do not need a computer science degree. You need the right questions." The FACTS Framework is designed to be used by anyone evaluating AI claims, regardless of technical background.

13. Most AI systems you interact with daily are invisible — they operate in the background without your awareness.

Answer **True.** The chapter describes AI as "a layer of software woven into the infrastructure of daily life — sorting, predicting, recommending, filtering, and deciding things on your behalf, often without your knowledge or consent." The Jordan example illustrates encountering six AI systems before 10 a.m. without being aware of any.

14. ContentGuard's moderation errors are randomly distributed across all user communities.

Answer **False.** The chapter states that ContentGuard's errors "are not random" and that "automated moderation systems disproportionately flag content from certain racial and linguistic communities while missing coded hate speech used by others."

15. Expert predictions for when general AI will be achieved are consistent and converge around a specific date.

Answer **False.** The Myth vs. Reality table states that "expert surveys show wide disagreement; median estimates range from 2040 to 'never.'"

Section 3: Short Answer (5 points each)

16. Explain the AI effect in 3–4 sentences. Include a specific historical example from the chapter.

Answer The AI effect is the human tendency to redefine intelligence to exclude whatever machines can already accomplish. When a machine achieves a task previously considered to require intelligence, people dismiss it as "not real AI." For example, when IBM's Deep Blue defeated chess champion Garry Kasparov in 1997, many people said chess-playing was "just brute-force computation" rather than real intelligence. Larry Tesler captured this by saying, "AI is whatever hasn't been done yet." This creates a moving-target problem that makes it difficult to accurately assess AI's current capabilities.

17. Describe the FACTS Framework and explain how each letter helps you evaluate an AI system. Use a specific AI example (not the depression-detection example from the chapter) to illustrate at least two of the five questions.

Answer The FACTS Framework consists of five questions: **F**unction (What task does this system perform?), **A**ccuracy (How well does it work, and for whom?), **C**onsequences (Who benefits and who might be harmed?), **T**raining (What data was it trained on?), and **S**tewardship (Who is responsible when it goes wrong?). For example, consider an AI resume-screening tool. **F**: It scans resumes and ranks candidates by predicted job fit. **A**: Does it work equally well for candidates from different educational backgrounds, countries, or who write in different styles? **C**: Employers benefit from faster screening; candidates from non-traditional backgrounds might be systematically disadvantaged. **T**: If trained on resumes of previously hired employees, it may replicate existing hiring biases. **S**: If a qualified candidate is unfairly rejected, is the employer, the AI vendor, or neither held accountable?

18. Why does the chapter argue that the narrow AI vs. general AI distinction is important for public understanding? Give one specific example of how conflating the two leads to confusion.

Answer The chapter argues this distinction matters because "almost all of the fear, hype, and confusion around AI comes from conflating the two." When people hear claims about AI's impact, they often unconsciously imagine general AI — a system that can do everything a human can — rather than the narrow AI systems that actually exist. For example, when someone warns that "AI will take all the jobs," they are implicitly imagining general AI. In reality, narrow AI automates specific tasks within jobs, which changes those jobs but does not simply eliminate them. Without the narrow/general distinction, people cannot accurately assess either the risks or the benefits of current AI systems.

19. Choose one of the four anchor systems (ContentGuard, MedAssist AI, Priya's Semester, or CityScope Predict) and explain how it illustrates the chapter's theme that "AI systems are tools built by humans with human biases, incentives, and blind spots."

Answer *Answers will vary. Strong answers will:* - **ContentGuard**: Show how the training data (posts labeled by human moderators) embeds the biases of those moderators — what they considered violations reflects their cultural assumptions. The design choice to prioritize speed over nuance reflects business incentives (scale and liability management). - **MedAssist AI**: Demonstrate how the training data (predominantly lighter-skinned patients in medical image datasets) reflects historical patterns in medical research that underrepresent certain populations. The over-reliance problem shows human psychology interacting with AI design. - **CityScope Predict**: Explain how historical crime data reflects decades of biased policing decisions — not objective measures of where crime occurs, but records of where police chose to look. The system automates and amplifies those human choices. - **Priya's Semester**: Discuss how AI tools are designed with particular use cases in mind (professional writing, general-purpose assistance) but deployed into educational contexts with different norms and expectations. The "blind spot" is the gap between the tool's design and the educational context.

Section 4: Applied Scenario (10 points)

20. Read the following scenario and respond to the prompts below.

A large online retailer announces a new AI-powered "personal shopping assistant" that analyzes your purchase history, browsing behavior, social media activity, and location data to predict what products you will want before you search for them. The company claims the system has a "95% satisfaction rate" based on internal testing. A tech blog headline reads: "Revolutionary AI Reads Your Mind to Find the Perfect Gift."

a) Apply the FACTS Framework to this system. Answer all five questions. (6 points)

b) Identify one way the headline is misleading and explain why. (2 points)

c) Is this system an example of narrow AI or general AI? Justify your answer. (2 points)

Answer **a) FACTS Framework:** - **F — Function:** The system analyzes user data (purchase history, browsing, social media, location) to predict and recommend products before the user searches for them. It does not "read minds" — it identifies statistical patterns in behavior data. - **A — Accuracy:** A "95% satisfaction rate" is vague. How was satisfaction measured? Did users rate recommendations, or did the company count any click as "satisfaction"? Does the system work equally well for users with different browsing habits, income levels, or from different geographic regions? What about new users with little data? - **C — Consequences:** The retailer benefits from increased sales. Users may benefit from convenient recommendations. However, users may be nudged toward unnecessary purchases (manipulative design). Users with less spending power might receive different quality of recommendations. The collection of social media and location data raises significant privacy concerns. - **T — Training:** The system was trained on user behavioral data, which reflects existing purchasing patterns. This data may underrepresent certain demographics, lead to stereotyped recommendations, or reflect historical inequities in access and marketing. - **S — Stewardship:** If the system uses social media data without clear consent, who is accountable? If it makes recommendations that lead to financial harm (e.g., encouraging overspending), who bears responsibility — the retailer, the AI developer, or the user? **b)** The headline "AI Reads Your Mind" is misleading because the system does not read minds — it analyzes behavioral data patterns. This anthropomorphizes the technology and implies a level of understanding and capability (mind-reading) that no current AI possesses. It conflates narrow pattern-matching with something resembling general intelligence or telepathy. **c)** This is narrow AI. It performs a specific task — predicting product preferences from behavioral data patterns. It cannot do anything outside this domain: it cannot have a conversation about why you want a gift, understand the emotional significance of a purchase, or adapt to a completely different task like medical diagnosis. Its versatility is limited to product recommendation within the retailer's ecosystem.

Scoring Guide

Score Performance Level
58–65 Excellent — strong grasp of all chapter concepts
49–57 Good — solid understanding with minor gaps
39–48 Satisfactory — core concepts understood, review recommended for weaker areas
26–38 Needs Improvement — significant review of chapter material recommended
0–25 Insufficient — re-read chapter and seek instructor support