What I tried, what worked, and how to try one piece of it in your course.
Weihao Qu · LearnAI Team · Department of Computer Science & Software Engineering
Left: raw particles, no instructions. Right: one structured prompt — paper, mountains, ensō, red seal. This whole deck was generated from that prompt. Hit I when you're ready to see it.
"Order and chaos dance — digital poetry in motion."
A structured prompt template — paper texture, ink-wash motifs, ensō circle, red seal placement, mist effects, Wabi-Sabi register. Drop it into GPT Image 2 (or Gemini 3 Pro Image) with your topic, get back any aesthetic you want for any course.
Best for: when "looks intentional" matters — research talks, public lectures, a deck you want students to remember by sight.
Here's the honest scope of what I shipped. Each tile clickable.
Claude Code plugin: drop a PPTX → get a clickable HTML lecture with Canvas visualizations + step-throughs.
Full courses, not just slides. Interactive automata + AI tutor, 12-module sandbox, PL playground, finance terminals — across 4 courses.
Pre/post survey across 8 courses (n=145 responses) · 153 multi-dept attitude survey · 17 controlled study · 6 interviews. MU IRB #2562.
Today I'll show you the slides + the research, plus a 4th piece that pulls it all together.
153 students across the university — CS, Bio, Chem, Math, Health, Psych, Business — told us how AI shows up in their college lives. Three findings that should change how we teach.
Multi-department survey · n = 153 students · CS, Bio, Chem, Math, Health, Psych, Business, others · MU IRB #2562 · Spring 2026.
Slide design is one piece. To build the interactive content itself — the clickable pointers, the simulators, the BF422 finance deck you just saw — pick from these two. Both turn topics or PPTX into self-contained HTML lectures.
Drop a PPTX, a PDF, or a topic. Get back a self-contained HTML deck with Canvas visualizations, step-through animations, and quizzes. This is the engine behind every demo today.
Best for: turning your existing lectures into interactive ones.
Describe a presentation in natural language. AI agents write React slide components on a standardized canvas, with hot reload, in-browser commenting, HTML/PDF export.
Best for: building from scratch with versioned, reviewable code.
All recipes documented at learnAIDoc. Free, open-source, install in minutes.
Same recipes from the previous slide → a complete AI literacy curriculum for non-CS undergraduates and high-schoolers. Every prompt I used is documented.
Claude Code (drafting) +
Codex (review) +
Whisper (transcripts).
161 wiki entries + 1 video transcript → 47 interactive HTML modules with terminal simulators, quizzes, XP tracking.
Free · open-source · MU IRB #2562 (amended) — pre/post measures + attitude surveys for non-CS learners.
mutest@monmouth.edu
test1234
Shared faculty test account — sign in to see the student experience.
12 modules · 15 interactive decks · 12 quizzes · sandbox · progress tracking. Built with the LearnAI undergraduate team.
Six volunteer CS205 students were interviewed at the end of Spring 2026 about studying with AI-enhanced slides + the course web app. Quotes below are anonymized; the percentage on each card is that student's final exam score, included to show the answers span the whole class — from 61% to 87%.
"AI just helped make 10 steps into animations — into one page."Final exam · 87%
"We always used to talk about it. If we didn't have the websites… it would be difficult."Final exam · 83%
"I'm a very visual learner. So it helps me to actually see the flow."Final exam · 78%
"You can play and see how each line works. I think that's very helpful."Final exam · 75%
"More like visuals — so I can compare and contrast."Final exam · 61%
Collected under MU IRB #2562 · 6 semi-structured interviews · CS205 Spring 2026 · anonymized · thematic coding (Braun & Clarke).
Pre-test 42% vs 41%. Both sections matched at baseline.
Treatment hit 80%, control 60% — on the same questions.
Small sample. Single course. Later-quiz analysis ongoing — ask in Q&A.
Collected under MU IRB #2562 · CS205 Spring 2026 · pre/post concept tests + 3 paired quizzes.
Pick a 30-minute slot — we walk through your discipline, your slides, your students. Free.
AI is the means. Your students are the end.
Thank you · monmouthaiteaching.com/research/slides/ for everything else (40+ decks, ATLAS, the dataset)
Open any of these during questions. Each is clickable in this deck or scannable from the booking slide.