Evidence Table

Master catalog of all artifacts cited in the supplementary chapters. Each artifact has a stable ID. Chapters cite by ID using inline parentheses, e.g., (WU-1.W01) or (WU-1.W01-Q1) for a specific verbatim quote.

The catalog is organized by iteration, then by artifact category within each iteration. Each iteration section opens with a brief overview of the iteration's institutional context and what its artifact corpus contains. Each category subsection opens with a summary of what kind of evidence the category provides for the chapters' analytic claims.

E.1 ID scheme

Format: {category}-{iteration}.{location} with optional -Q{n} for quote-level citation.

E.2 Iteration 1 · CTD Pilot · Spring 2024

The first iteration ran from January 17 through May 1, 2024, as the CTD pilot at the ATLAS Institute with approximately twenty-five undergraduate students. This is the discovery phase of my pioneer practice: the curriculum was being built from scratch, the four-theme architecture was being stabilized through delivery, and my reflective journaling was the densest contemporaneous record of any iteration. The Iteration 1 corpus is the most analytically valuable single iteration's data in the dissertation because it carries both the curricular structure as students experienced it and my week-by-week analytic reflection on what the teaching was teaching me.

E.2.1 Curricular artifacts

The curricular corpus for Iteration 1 documents what was taught, in what sequence, with what materials. It includes the full Canvas LMS export, the course-framing deck that established the four-theme architecture for the iteration, the Week 1 opening lecture with twenty-three named student outputs, and the Final Project assignment that operationalized the four themes through a Media Studies-style company-creation task.

CV-1 · Canvas LMS export, Spring 2024. 225 files; 15-week modules including 4 video recordings of guest lectures.

DK-1.FG · ForeverGold course deck, Spring 2024. 36 slides; the four-theme architecture (Education, Industry, Ethics, Accessibility) appears on slides 5 through 9.

DK-1.W01 · Week 1 opening lecture, January 17, 2024. 35 slides; 23 named student prompt-engineering outputs, including Ashley Stafford and Ethan Cuenca whose work persists across subsequent iterations.

FP-1 · Final Project Requirements, Spring 2024. Multi-tool Media Studies-style company-creation assignment with 9 reflection questions distributed across the four themes, including a dedicated ethics question on creative-job displacement.

E.2.2 Reflective artifacts

The reflective corpus for Iteration 1 is uniquely well-developed in the dissertation. The Weekly Updates Prelim Document is the only contemporaneous structured reflective journal in the corpus, kept across fifteen weeks with a consistent format ("Curriculum Development / Special Guests / Resources / Learned / Students Produced") per week. The "Learned" sections function as analytic reflexivity in Anderson's strict sense: each week records what the teaching had taught me, captured at the time the teaching happened. This is the densest contemporaneous reflective record any iteration produced.

WU-1.W01 · Weekly Update Week 1 (lecture date January 17, 2024). My first structured "Learned" reflection.

• WU-1.W01-Q1: "I learned that some of the multiple choice quizzes generated by ChatGPT were not correct and had hallucinations."
• WU-1.W01-Q2: "I learned that the students enjoyed creating the alter-ego images to describe themselves to the class."

WU-1.W02 through WU-1.W15 · Selected "Learned" entries from Weeks 2 through 15. Cited in Chapter C §C.2.8 as the arc-anchoring quotes across the semester. Full text in the Weekly Updates Prelim Document.

• WU-1.W02-Q1: Week 2 entry on Midjourney access and image-generation bias
• WU-1.W03-Q1: Week 3 entry on AI content detectors and institutional GenAI policy
• WU-1.W05-Q1: Week 5 entry on cross-industry GenAI application
• WU-1.W06-Q1: Week 6 entry on SORA and GenAI in finance
• WU-1.W07-Q1: Week 7 entry on deepfake ethics and music generation
• WU-1.W14-Q1: Week 14/15 entry on assistive technologies and Final Project execution

E.3 Iteration 2 · Mixed Engineering · Spring 2025

The second iteration ran from January 13 through April 30, 2025, as GEEN 3830-001 Special Topics in the College of Engineering and Applied Science with approximately twenty-five undergraduates of mixed engineering backgrounds. This is the consolidation phase of my pioneer practice: the four-theme architecture had stabilized through Iteration 1, the tool landscape had shifted substantially (DeepSeek, AI Agents, NotebookLM, Claude 3.7 all entered the curriculum), and the iteration introduced the student teach-out as a new pedagogical move. The Iteration 2 corpus is the strongest single source for the tool-turnover and module-placement-evolution patterns developed in Chapter C §C.6.

E.3.1 Curricular artifacts

The curricular corpus for Iteration 2 documents the consolidated mid-iteration form of the curriculum: a syllabus that names the four-theme architecture explicitly with weekly structure, the Week 1 prompt-engineering opener that preserved Iteration 1's structural commitment, and five iteration-specific lecture decks covering the new tools and themes that emerged between Spring 2024 and Spring 2025.

CV-2 · Canvas LMS export, Spring 2025. 240 files; 15 weeks of modules; 103 HTML pages; 6 video recordings of guest lectures.

SY-2 · Syllabus for GEEN 3830-001 Special Topics. Three-credit course; classroom DLC170; weekly structure naming the four themes; tool list including DeepSeek, Claude 3.7, Grok 3, Sora, HeyGen, Be My AI, ElevenLabs.

DK-2.JAN13 · Week 1 Prompt Engineering opener (January 13, 2025).

DK-2.JAN15 · Midjourney Self-Portrait Assignment deck (January 15, 2025).

DK-2.FEB05 · Future Wheel technique deck (February 5, 2025).

DK-2.MAR05 · Industry theme deck (March 5, 2025).

DK-2.MAR10 · Education theme deck (March 10, 2025).

E.3.2 Learner-facing artifacts

The learner-facing corpus for Iteration 2 includes student teach-out presentations, a new pedagogical move I introduced in this iteration. The teach-outs gave students an opportunity to research a topic of their choice and present it to the class, generating six recorded presentations preserved in the archive. The teach-out topics are distributed across the four themes and provide concrete evidence for the dialogue-with-informants-beyond-self criterion (§B.3.4).

SP-2.DEEPFAKE · Student teach-out: DeepfakeAI (Daniel Debretsion). Ethics topic with Industry implications.

SP-2.SINTRA · Student teach-out: Sintra.ai (Alt Style). Industry topic on AI-driven business workflow tools.

SP-2.HALEY · Student teach-out: Haley Phillips (two files).

SP-2.DAKOTA · Student teach-out: Dakota A.

SP-2.ROBOTICS · Student teach-out: The Integration of AI and Robotics. Industry topic on multi-modal AI in robotics.

E.3.3 Administrative artifacts

The administrative corpus for Iteration 2 documents the iteration's operational organization: the guest-lecturer rotation, the teach-out scheduling, and the final-presentation order. These spreadsheets are the operational trace of the curation work that Chapter C §C.6.3 develops as a pedagogical contribution in its own right; they show how the guest network and the student presentations were assembled into a coherent semester structure.

AS-2.GUESTS · Guest lecturer list spreadsheet. Records Tom Yeh (DeepSeek), Nolan Brady (NeuroImaging), Bobby Hodgkinson (NotebookLM), Justin Shacklette (GenAI for Software) as the iteration's central guests.

AS-2.TEACHOUT-DATES · Teach-out dates spreadsheet.

AS-2.TEACHOUT-INDEX · Teach-out slides index.

AS-2.FINAL-SCHED · Final presentations schedule.

E.4 Iteration 3 · GenAI in Five Cohort 1 · Aug 2025

The third iteration was the first compression of the curriculum from a fifteen-week semester to a five-day online workshop, running August 18 through August 22, 2025, sponsored by the College of Engineering and Applied Science (CEAS) and hosted on Luma. This is the first phase of distillation in my pioneer practice. The Iteration 3 corpus is the strongest single source for the workshop-format viability claim and for the learner-facing feedback that supports the compression-as-curriculum-maturation finding (§D.3).

E.4.1 Curricular artifacts

The curricular corpus for Iteration 3 is the workshop deck that compressed the full semester's content into a 22-slide, five-day structure with one topic per day (Monday image, Tuesday video, Wednesday audio, Thursday research, Friday human-centered AI).

DK-3 · Workshop deck, 22 slides; one topic per day across Monday through Friday.

E.4.2 Learner-facing artifacts

The learner-facing corpus for Iteration 3 is the principal triangulation data for the iteration. The Luma feedback survey produced twenty-nine responses with nine carrying text comments; the Luma participant roster documents the audience composition with unusual specificity (411 registered, 129 attended live, 65% students, 70% Master's-interested). Together, these data sources support the iteration's pedagogical-viability claim and surface the depth-versus-breadth tension developed in Chapter C §C.4.6.

LF-3 · Luma feedback aggregate. 29 responses; 22 fives, 6 fours, 1 two; average 4.69 of 5; 9 responses with text comments.

• LF-3-Q1 (5★): "Best"
• LF-3-Q2 (5★): "I loved it!"
• LF-3-Q3 (5★): "I was wonderful session"
• LF-3-Q4 (5★): "It was really good"
• LF-3-Q5 (5★): "Excellent"
• LF-3-Q6 (5★): "Good many tools explored and learned now I'm trying one by one few tried"
• LF-3-Q7 (5★): "I really enjoyed the session by Larissa Schwartz, where she introduced us to the latest AI image generation tools like Midjourney, Microsoft Designer, DALL·E, Canva, Adobe Firefly, and NightCafe. She..." (response truncated in source)
• LF-3-Q8 (4★): "Very interesting."
• LF-3-Q9 (4★): "I think it was a good discussion regarding how to use the different AI image generation tools. A course work based on some of the neural networks behind them could be a great one."

LF-3.R01 through LF-3.R29 · Individual Luma feedback responses 1 through 29.

LR-3 · Luma roster. 411 registered over the 8-month registration window; 129 attended live; 65% students; 70% expressed interest in an AI Master's program.

E.5 Iteration 4 · GenAI Works Cohort · Sept 2025

The fourth iteration was the second compressed cohort, delivered September 8 through September 12, 2025, through partnership with GenAI Works and broadcast on the GenAI Works YouTube channel. This is the second phase of distillation in my pioneer practice and the iteration with the most extensive recorded teaching delivery in the corpus. The Iteration 4 corpus is the strongest single source for the hallucination-as-pedagogy finding (§D.2) and for the multi-channel teaching practice finding (§D.4) at scale.

E.5.1 Curricular artifacts

The curricular corpus for Iteration 4 is the same workshop deck template as Iteration 3, with date headers updated to reflect the September schedule. The template stability between cohorts is itself evidence for the compression-as-curriculum-maturation finding.

DK-4 · Workshop deck, 22 slides; same template as DK-3 with date headers updated to September 8 through 12, 2025.

E.5.2 Audience and feedback

The Iteration 4 audience data is substantially larger than Iteration 3's, reflecting the GenAI Works partnership's broader reach. The Iteration 4 Luma event page (https://luma.com/m31ygqao) accumulated registrations across the cohort's promotion period, and the YouTube live broadcasts captured real-time participation numbers per session.

LR-4 · Iteration 4 Luma roster. Approximately 4,731 registered guests on the Luma event page across the cohort's registration window. The Iteration 4 partnership with GenAI Works (a community of approximately six million LinkedIn followers per Day 1 transcript reference) drove substantially higher registration than the CEAS-sponsored Cohort 1.

LR-4.D1 · Iteration 4 Day 1 live participants. 2,654 online participants joined the Day 1 YouTube broadcast on September 8, 2025.

LF-4 · Iteration 4 Luma feedback aggregate. 256 ratings; average 4.2 of 5.

E.5.3 Delivery artifacts

The delivery corpus for Iteration 4 is the largest single text-volume source in the dissertation. The five day-by-day YouTube transcripts together total approximately 55,000 words of contemporaneous teaching delivery, with Day 1 carrying the hallucination teaching passage that supports the most recent anchor for the hallucination-as-pedagogy finding.

TR-4.D1 · Day 1 transcript: Image generation. 10,962 words. Tom Yeh guest from CU Boulder; real-time global audience visible (Nigeria, UK, Denver, Costa Rica named in the first 30 lines).

• TR-4.D1-Q1: Opening session, host welcoming attendees and naming the geographies in the chat ("Amazing. Welcome everybody... We got um people from all over the world. I see Nigeria, Denver, UK, Costa Rica.")
• TR-4.D1-Q2: Host introducing the workshop's co-hosts ("So, uh we have Tom Yei. welcome from uh you know University of Colorado uh Boulder. We have Lissa Schwarz as well is going to be our main host today.")
• TR-4.D1-Q3: Hallucination teaching passage during Microsoft Designer walkthrough ("And you can also see these different hallucinations that are going on. So there aren't even bodies in these shoes... You can keep inputting and inputting and inputting the prompts...")

TR-4.D2 · Day 2 transcript: Video generation. 11,395 words.

TR-4.D3 · Day 3 transcript: Audio generation. 11,599 words.

TR-4.D4 · Day 4 transcript: Research tools. 9,921 words.

TR-4.D5 · Day 5 transcript: Human-Centered AI and Vibe Coding. 11,276 words.

E.6 Cross-iteration evidence · 2023 to 2026

The cross-iteration corpus contains the artifacts that span more than one iteration or sit outside the iteration framework entirely. They include the institutional and program context that situates the entire research arc, my public-facing reflective writing across multiple channels, K-12 outreach work, external media coverage, and federal-research engagement. The cross-iteration evidence is what makes the multi-channel teaching practice finding (§D.4) documentable: it shows the four-theme architecture operating across eight delivery channels and three distinct audience age ranges over more than two and a half years.

E.6.1 Program and grant context

The program and grant artifacts situate the work within its institutional setting. The AI-IRT Seed Grant proposal documents the seeded research arc that connects to both committee members (Tom Yeh and Diane Sieber as co-PIs), and the ENED Preliminary Exam Part 2 records my commitment to offering a continuous generative-AI course at CU Boulder.

AI-PROPOSAL · AI-IRT Seed Grant proposal. Principal investigators Tom Yeh (Computer Science) and Diane Sieber (Herbst Program), both on my committee. Explicit K-12, undergraduate, and graduate scope.

PR-PART2 · ENED Preliminary Exam Part 2, Innovation Proposal. Documents my intent to offer a continuous Generative AI class for CU Boulder because there had been only one such class offered previously (the one I had taught).

E.6.2 Reflective writing

The cross-iteration reflective writing carries my first-person voice across the 2024-2026 period and is the principal supplementary source for the analytic-reflexivity criterion in Iterations 2 through 4 (where the structured weekly journaling of Iteration 1 has no direct counterpart). The Research Impact Essay anchors my practitioner-pioneer biography; the Keep Up Newsletter publishes monthly tool-focused reflections; the Keep Up Podcast extends the same content into audio form; together they provide the public-facing reflective channel that complements the iteration-internal data.

RE · Research Impact Essay, c. Spring 2024. Three-paragraph self-narrative written in the first person, likely for a fellowship application.

• RE-Q1: "Over a year ago I started researching Generative AI and how teachers and students can use different art applications, such as DALL E, Midjourney and NightCafe in order to create artwork from a prompt."
• RE-Q2: "I shared my ideas with a school district in Colorado, where I led a professional development to teach the staff how to incorporate and collaborate with these tools in the classroom."
• RE-Q3: "From this art contest, we had over 60 students compete and it was such a success that it motivated me to pilot the first ever Generative AI class at the University of Colorado Boulder through the Atlas Institute."
• RE-Q4: "During this spring semester, I've been teaching Generative AI to undergraduate students about how these different applications can be used in Education, Industry and Accessibility, and the ethical concerns that arise while using these programs."

KN-EP1 · Keep Up Newsletter Episode 1, Image Generation (April 24, 2025). Public LinkedIn Pulse newsletter; first episode using the running-and-training metaphor.

• KN-EP1-Q1: "Expect variable results, occasional hallucinations; persistence improves prompting skills; join community groups" (Lessons-from-Training section).

KN-EP2 · Keep Up Newsletter Episode 2, Research Tools (May 7, 2025).

KN-EP3 · Keep Up Newsletter Episode 3, Sound Tools (May 23, 2025). Publicly credits an Iteration 1 student for introducing Soundful into the curriculum.

• KN-EP3-Q1: "I learned about Soundful from one of my students during class." (the Ethan Cuenca → Soundful student-to-instructor tool flow)

KP-EP2 · Keep Up Podcast Episode 2 transcript, May 2025. ~30 minutes; AI by Hand YouTube channel.

KP-EP3 · Keep Up Podcast Episode 3 transcript, May 2025.

E.6.3 HCI summer 2024 guest series

The HCI guest series consists of ten lectures I delivered for a Human-Computer Interaction course at CU Boulder in summer 2024. The series is the most direct transport of the four-theme curriculum architecture into a different course taught by a different instructor; it is the primary evidence for the curriculum-portability claim (§A.5) and contributes to the multi-channel teaching practice finding.

HC-INTRO · HCI Generative AI Intro and Prompt Engineering deck.

HC-ACCESS · HCI Generative AI within Accessibility.

HC-AUDIO · HCI Generative AI within Audio (Ethics).

HC-EDU · HCI Generative AI within Education.

HC-INDUSTRY · HCI Generative AI within Industry.

HC-MUSIC · HCI Generative AI within Music (Ethics).

HC-VIDEO-ETH · HCI Generative AI within Video (Ethics).

HC-VIDEO · HCI Generative AI within Video.

HC-FUTURE-WHEEL · HCI Prompt Engineering using Future Wheel.

HC-SORA · HCI Key Findings in Relation to Sora.

E.6.4 K-12 outreach and research

The K-12 outreach corpus documents my work with children and teens at the CU Boulder STEAM Festival and through the University of Washington KidsTeam and Youth Advisory Board (YAB) co-design sessions in summer 2024. The STEAM Festival mural was a live workshop activity I led; the UW KidsTeam collaboration was a co-design research project I joined as a participant. Together they provide the K-12 end of the audience age range and contribute the near-independent corroboration for the hallucination-as-pedagogy finding (KT-THEMES-C5).

ST-MURAL · AI Art Mural with 50+ Colorado children at the CU STEAM Festival, 2024. IDC conference submission document.

ST-PHOTO · STEAM Festival workshop photograph showing the mural being built live.

KT-DECK · UW KidsTeam Spencer Project slide deck.

KT-IDEAS · UW KidsTeam 2024 three-day AI Ideas session plan.

KT-COMIC · KidsTeam Comic Boarding worksheets (art and math AI-in-schools scenarios).

KT-NOTES · KidsTeam observational and analytic notes on session videos.

KT-YAB · UW Youth Advisory Board (YAB) Multimodal-AI session deck.

KT-THEMES · KidsTeam Research Questions and surfaced Themes, July 2024. Children's and teens' independent observations on GenAI in schools, surfaced through the UW KidsTeam co-design methodology (which the UW team led; I participated).

• KT-THEMES-C1 (challenges): GenAI being banned from schools
• KT-THEMES-C2 (challenges): Cheating
• KT-THEMES-C3 (challenges): AI as "data fed into a computer" that "doesn't yet know everything"
• KT-THEMES-C4 (challenges): Lack of emotion and human connection
• KT-THEMES-C5 (challenges): "Hallucinations (such as images produced with a third arm)"
• KT-THEMES-O1 (opportunities): Teachers allowing GenAI for paragraphs with mistakes to correct
• KT-THEMES-O2 (opportunities): Checking math homework
• KT-THEMES-O3 (opportunities): Producing faster work

E.6.5 External media coverage

External media coverage documents how journalists framed the work for non-academic audiences. The single article in the corpus, from Aspen Public Radio, appeared one week after Iteration 1 ended and provides external corroboration that the iteration landed pedagogically.

AP-2024-05-16 · Aspen Public Radio article, May 16, 2024. Title: "Could AI be the next college teaching assistant? Some Colorado professors believe so." Larissa named and quoted; Iteration 1 student Ashley Stafford also quoted. Cited at source-level.

E.6.6 Federal research engagement

The federal-research engagement is the most recent public-facing synthesis of the work, delivered approximately ten weeks before defense. The CU RMACC webinar, uploaded by the federal NAIRR Pilot program, makes the four-theme architecture and the pioneer practice visible to a federal-research audience.

WB-2026-03-03 · CU RMACC Webinar via the federal NAIRR Pilot platform, March 3, 2026. "Unleashing Creativity with Generative AI"; 56 minutes; ~9,600 words of transcribed delivery. Cited at source-level throughout this dissertation.

E.6.7 Other public-facing artifacts

A small number of additional public-facing artifacts complete the cross-iteration corpus. They document teaching activities that did not fit into the main iteration framework but contribute to the multi-channel teaching practice claim.

DK-DLS · DLS Prompt Engineering presentation. Undated; 17 slides; four named participant outputs. Delivered at the CU Discovery Learning Center.

STC · Storytelling Cartoonimator worksheet. K-12-oriented worksheet pairing storytelling with the Cartoonimator tool.

E.7 Conventions

• Verbatim quotes preserve original punctuation. Minor clarifications use [sic] only when needed for reader comprehension.
• Q-IDs are assigned on demand as chapters cite. Adding a quote that does not yet have an ID is a normal step during drafting.
• New artifacts added after 2026-05-12 receive IDs at ingestion. The category prefix table above is extensible.
• Multiple citations in one parenthesis: separate with semicolons, e.g., (WU-1.W01-Q1; KT-THEMES-C5; KN-EP1-Q1).