The Teller
Courses and Programs · Case study
Every student builds the same kind of storyteller, the Teller, in Gemini Notebook. What it knows, how it speaks and how it looks are theirs to design.
What a Teller Is
A Teller is a Gemini Notebook loaded with one narrator’s consented archive: the interview transcript, photographs, documents and the student’s field journal. It answers only from those sources, with citations, and follows instructions the student writes. The student interviews it on the record, and the edited recording is the finished story.
Students choose one of three story paths. On the main path, a real narrator signs the consent package below. A student may instead interview themselves, with no consent package, or build a Teller for a character or icon (a fictional character, a brand figure, a place or object, or a historical figure in the public domain) from public sources, with a source list and a credit line in place of consent. No living real person ever speaks in the first person, and every path is graded the same.
Two Modes
| Mode | How it speaks | Consent needed |
|---|---|---|
| Narrator | A voice that knows the archive and speaks about the person, quoting their words | Signed Agreement item 5 |
| First person | Speaks as the person, quoting the archive, like the grandfather bot in Artificial | Item 6 for a living narrator; Part D, signed by a family member or estate holder, for someone who has died |
Either mode is graded the same. In both, the Teller’s first answer says it is an AI built from an archive.
Tools by Stage
Gemini and Adobe Creative Cloud come through the college, and free tools are welcome. Recording and editing need a Windows or macOS computer that runs Premiere Pro; campus computer labs are open to every student.
Students without a computer or reliable internet can borrow a laptop or hotspot through GCC’s Office of Information Technology (OIT) Student Technology Request.
| Stage | Google and AI | Adobe | Free and other |
|---|---|---|---|
| Pitch and interview | Gemini app for loglines and questions; Gemini Live for rehearsal | Audition | Phone voice memo |
| Archive | Drive and Sheets | Premiere Pro speech-to-text, Photoshop | Phone scanning |
| Building the Teller | Gemini Notebook | ||
| Voice guide and audit | Google Docs and Sheets | ||
| Conversation map | Google Slides | InDesign, Illustrator | Paper thumbnails |
| Visual identity and key art | Photoshop, Illustrator, Express | Krita, pen and paper | |
| The Teller’s voice | Gemini Notebook audio features, labeled as AI | Audition, Enhance Speech | Audacity |
| Recording the interview | OBS Studio, Meet or Zoom | ||
| Editing the story | Premiere Pro (text-based editing, captions) | DaVinci Resolve | |
| Case study | Google Sites | Adobe Portfolio |
Consent Options
On the real narrator path, the narrator answers yes or no to each use in the Signed Agreement. Any item left blank is no.
| Option | What the narrator agrees to |
|---|---|
| AI processing (item 4) | The recording, transcript, photographs and documents may be processed by Google Gemini and Adobe tools through the student’s Maricopa accounts. |
| Narrator mode (item 5) | A Teller may answer questions about the narrator, quoting their words. |
| First-person mode (item 6, or Part D) | A Teller may speak as the narrator in the first person, quoting their recorded words. Default no. |
| Labeled synthetic voice (item 13) | A synthetic AI voice may read the Teller’s answers in the recording, labeled as AI on screen and in the credits. It is never made to sound like the narrator. Default no. |
| Withdrawal (Part E) | “If you change your mind, I’ll delete your material within seven days and confirm in writing.” |
Rules for Every Teller
- It says it is an AI. In every mode, and on screen in the recording.
- Every visual is authored. The Teller’s portrait, archive cards and key art come from the archive and the student’s own drawing and photography. Gemini writes; it never designs or generates images or video.
- No cloned voice or likeness. A synthetic voice is labeled as AI and never imitates a real person; the Teller’s face is never a likeness of the narrator.
- Answers are never changed. The edit may shorten the interview, but not what the Teller said, and every cut inside an answer is marked.
- Consent travels with the material. Nothing from an interview goes into any AI tool until the narrator signs the Signed Agreement that allows it.
- The Teller is never public. It is shown live only inside the class, and otherwise by recording.
- Everything is disclosed. Every process step carries the AI-use log, and the final carries a credit line naming what was made by hand and what was assisted.