Exercises: AI and Creativity

Part A: Foundational (Understand/Apply)

Exercise 11.1 — Generative Model Comparison

In your own words (no more than 100 words each), explain how each of the following generative approaches works. Use an analogy or metaphor that someone without a technical background would understand.

1. Autoregressive text generation (LLMs)
2. Generative Adversarial Networks (GANs)
3. Diffusion models

Then answer: What do all three approaches have in common?

Exercise 11.2 — Locating on the Authorship Gradient

For each of the following scenarios, locate the work on the authorship gradient (from "fully AI-generated" to "fully human-created") and explain your reasoning in 2–3 sentences:

1. A musician hums a melody, uses AI to generate a full orchestral arrangement, then adjusts the instrumentation.
2. A novelist writes an entire book but uses AI to generate chapter title suggestions.
3. A marketing team has AI generate 100 ad slogans, selects the best five, and uses them verbatim.
4. A student asks an LLM to "write a 500-word essay on climate change" and submits the output unedited.
5. An architect uses AI to generate 3D renderings of their hand-drawn floor plans.

Exercise 11.3 — Vocabulary in Context

Choose five key terms from this chapter and write a short scenario (3–4 sentences each) showing how the term applies in a real-world situation. Do not use examples from the textbook.

Exercise 11.4 — The Creativity Test

Look at three AI-generated images (you can find them on any text-to-image tool, or use screenshots from published examples). For each image, evaluate:

1. Is it novel — does it show something you haven't seen before?
2. Is it skilled — does it demonstrate technical competence?
3. Is it expressive — does it seem to communicate something?
4. Is it creative? Why or why not?

How did your definition of "creative" shape your answers?

Part B: Intermediate (Apply/Analyze)

Exercise 11.5 — The Training Data Audit

Choose a generative AI tool you have access to (a text generator, image generator, or music generator). Try to answer the following questions by reading the tool's documentation, terms of service, and any published information:

1. What data was the model trained on?
2. Were the original creators of the training data compensated?
3. Can creators opt out of having their work used for training?
4. Does the company claim any rights over the content you generate with the tool?
5. What gaps or limitations did you encounter in trying to answer these questions?

Write a 300–400 word summary of your findings and what they reveal about the training data provenance problem.

Exercise 11.6 — Perspective-Taking: Three Stakeholders

Consider the scenario of a company that develops an AI music generator trained on the work of independent musicians. Write a short statement (150–200 words) from the perspective of each of the following stakeholders:

1. The AI company's CEO: Argue why the tool benefits society.
2. An independent musician whose work was used in training without consent: Explain your concerns.
3. A small business owner who uses the tool to generate background music for their YouTube channel: Explain why you find the tool valuable.

After writing all three, write a 100-word reflection on which perspective you found most compelling and why.

Exercise 11.7 — Copyright Scenario Analysis

Read the following scenario and analyze it using the frameworks from Section 11.4:

A graphic novelist spends three months creating a 200-page graphic novel. She writes all the dialogue and narrates the plot herself. For the visual panels, she uses an AI image generator, crafting detailed prompts for each panel, generating dozens of options per panel, selecting the best versions, and editing them in Photoshop to maintain visual consistency. She estimates the AI generated the raw images, but she spent roughly 400 hours on prompt crafting, selection, editing, and layout.

Questions: 1. Should this work be copyrightable? Under current U.S. rules, which elements would be protected? 2. Where does this work fall on the authorship gradient? 3. If the AI model was trained on the work of other graphic novelists (without their consent), does that change your analysis? 4. Compare this to a photographer who uses Photoshop's AI-powered tools extensively. Is there a meaningful difference?

Exercise 11.8 — Displacement vs. Democratization Mapping

Choose a creative industry (illustration, music, writing, photography, film, graphic design, etc.) and create a two-column analysis:

After completing the table, answer: 1. Are the people being displaced and the people being empowered the same people? 2. What policies could address the displacement while preserving the democratization? 3. On balance, is AI a net positive or net negative for this industry? Explain your reasoning.

Part C: Advanced (Evaluate/Create)

Exercise 11.9 — Policy Proposal: Fair Compensation for Training Data

Several proposals have been suggested for compensating creators whose work is used to train generative AI models. Research or reason through the following options and evaluate each:

1. Individual opt-in licensing: AI companies must get permission from each creator before using their work.
2. Collective licensing fund: AI companies pay into a fund that is distributed to creators proportionally.
3. Opt-out registry: Creators can register to have their work excluded from training datasets.
4. No compensation needed: Training on publicly available data is fair use and requires no compensation.
5. Royalty on output: Creators receive a small royalty whenever AI-generated content is similar to their work.

For each option, assess: feasibility, fairness to creators, impact on AI development, and enforceability. Then state which option (or combination) you would recommend, and why.

Exercise 11.10 — The Creativity Debate Essay

Write a 600–800 word argumentative essay taking a position on the question: "Can AI be creative?" Your essay should:

1. State your position clearly in the first paragraph.
2. Define what you mean by "creativity" — and acknowledge that your definition shapes your conclusion.
3. Engage with at least two of the three philosophical perspectives discussed in Section 11.3.
4. Use specific examples of AI-generated content to support your argument.
5. Acknowledge the strongest counterargument to your position and respond to it.

Exercise 11.11 — Human-AI Creative Collaboration Experiment

Complete a creative project using AI as a collaborator. You may write a short story, create a series of images, compose a song, design a poster, or any other creative work. Document the process:

1. What was your creative goal or vision?
2. How did you use the AI tool? What prompts or inputs did you provide?
3. What did the AI contribute? What did you contribute?
4. Where on the authorship gradient would you place the final result?
5. How did the experience compare to your usual creative process? Was the result "yours"?
6. What did the experience teach you about the nature of creativity?

Submit both the creative work and the process documentation.

Exercise 11.12 — Industry Impact Scenario

It's 2030. Choose a creative industry and write a 500–700 word scenario describing what the industry looks like after five more years of generative AI development. Your scenario should be realistic (not utopian or dystopian) and should address:

1. Which roles have been most affected?
2. What new roles have emerged?
3. How has the business model changed?
4. How have copyright and compensation issues been resolved (or not)?
5. What's the relationship between human and AI creativity in this industry?

Reflection Question

Think about a piece of art, music, writing, or creative work that has been meaningful to you personally — something that moved you, changed your perspective, or stuck with you over time. Could an AI have created something that would have affected you in the same way? Why or why not? What does your answer tell you about what you value in creative work?