Estimated time: 75 minutes. Competencies: AI ethics literacy, AI feedback evaluation.
make_videos.py or record and embed your own.)By the end of this lesson you will be able to:
Think about the song you made in Lesson 1.1. List everyone and everything that contributed to it: you, the model, the people whose recordings trained the model, the company that built it, the tool that mixed it. Now ask: who should be credited, who should be paid, and who was asked? Keep the list; you will return to it.
1. Training data and consent. Text-to-music models learn from large collections of recordings. In many cases the artists whose work was used were not asked and were not paid. Some companies have licensing deals; others rely on arguments about fair use that courts are still testing. As an educator, you do not have to settle the law, but you do have to be honest with students that the tool's fluency rests on other people's work.
2. Voice and likeness. Models can imitate a recognizable singer's voice. Several jurisdictions treat a person's voice as part of their right of publicity, and some have passed laws specifically about AI voice cloning. The course rule is simple: do not clone the voice or signature style of a real, named artist. Prompt for a type of voice ("husky tenor"), not a person.
3. Ownership and copyright. In the United States, the Copyright Office has stated that works generated entirely by AI without human authorship are not eligible for copyright, while works with meaningful human creative contribution can be, to the extent of that contribution. That means your lyrics and your arrangement choices may be protectable; the raw generation may not be. Tools' terms of service also differ on who may use outputs commercially, and free tiers often restrict this. Read the terms before you release anything.
4. Disclosure. Audiences, students, and collaborators deserve to know when AI was involved. Disclosure protects trust and models good practice. It is the reason for the one-line credit under every artifact in this course.
When you analyze a case, work through these steps and write a sentence for each:
| Step | Ask |
|---|---|
| Stakeholders | Who is affected: artists, listeners, students, companies, the public? |
| Harms and benefits | What does each stakeholder gain or lose? |
| Consent | Was permission sought from the people whose work or likeness was used? |
| Transparency | Was the AI role disclosed? |
| Alternatives | Could the same goal have been reached with less harm? |
| Position | What should have happened, and what rule would you set in your classroom? |
Later in this lesson you will ask a text model to summarize a legal question. Expect a confident, fluent answer that may be out of date or simply wrong. Models are trained on text up to a cutoff, they blend jurisdictions, and they sometimes invent case names. Treat the answer as a first draft to verify, and note every claim you cannot confirm from a primary source (a court filing, a government office, the tool's own terms).
"Lyrics written by [name]. Melody, arrangement, and vocals generated with [tool and version] from those lyrics and a style prompt written by [name]. Final mix edited by [name] in [tool]. This work includes AI-generated material and is shared for educational use."
Lesson-1-4_LastName_Ethics.pdf.Post in the Lesson 1.4 Forum:
Then reply to two peers who took a different position or analyzed a different case. Engage with their reasoning, not just their conclusion.