The Glass Syllabus· AI in Academia
Presenter view Projector waiting June 3 Agenda
A working guide for faculty

AI in Academia:
What is the pedagogy approach?

Take a stance first, then we listen, then we build. By the end, we revisit your stance to see what moved.

60-minute sessionFor facultyVote, listen, compare
Press or the Next button to advance.
Welcome

Today's working hour on AI in the classroom.

Vote your gut reaction first. Listen to a five-minute podcast with hidden prompts. Work through five strategies. Revisit the vote at the end to see what moved.

AI in academia · the pedagogy approach

Three parts in 60 minutes

1

The Instructor's Role

The Role

What is the instructor's role in the age of AI?

The Examples

Which of your own assignments live up to that role?

Responses stay private. Common themes are revealed when we reach the instructor section.

2

The Student's Role

The Role

What are your thoughts on the student's role?

The Examples

Which student behaviors show that role in action?

Themes from this card are revealed when we reach the student section.

3

The Role of AI

The Role

What is AI's role in your teaching and your students' learning?

The Examples

Which tools would you use, and what would you use them for?

Combined with the room's list when we reach the section on AI.

Live: 0 responses from 0 in the room
First, your gut

Ban it, ignore it, or embrace it?

No coaching. No nuance yet. Your honest gut reaction right now. We are saving this number for the end of the hour.

Your honest gut reaction to students using AI?

Tap an option to cast a vote.

🚫 Ban it0
🙈 Ignore it0
🎉 Embrace it0
Total votes: 0
Listen now

A podcast on AI in pedagogy

Five minutes. NotebookLM hosts. Listen for the four prompts they slip in.

The rule: just listen. The hosts will drop four things you have to do. Catch them. We will share after.
🎧

Your NotebookLM podcast goes here

Auto-loaded from /videos/podcast-main.m4a. Use the button above to swap in a different file on the fly.

Now flip the cards

The four exercises from the podcast

Click a card to flip. Tap a colored ball for your call. We reveal the room's picks at the end.

1
Exercise One
Click to reveal
Color the example

Graduate students must write a 1000-word literature review of recent research on climate adaptation policy.

What color would you give it?

Pick:
Green0%
Yellow0%
Red0%
2
Exercise Two
Click to reveal
Color the example

Students design a community engagement plan and identify three stakeholders to interview.

What color would you give it?

Pick:
Green0%
Yellow0%
Red0%
3
Exercise Three
Click to reveal
Your own assignment

Pick one assignment from your own class right now. Just one. Name it.

What color is your assignment right now?

Pick:
Green0%
Yellow0%
Red0%
4
Exercise Four
Click to reveal
Redesign it

Look at the assignment you just named. What is one of the smallest changes you could make that would either shift its color or make it harder for AI to hand the answer over?

What color is it after your redesign?

Pick:
Green0%
Yellow0%
Red0%
Now share for 90 seconds: turn to one person near you and read your exercise three and your exercise four. Then we pick two to share with the room.
The framework

The Glass Syllabus AI Traffic Light

A framework instructors use to label each assignment with one color, signaling exactly how much AI is allowed on that specific task. The point: stop treating AI as a single yes-or-no rule for the whole course. Start deciding per assignment, on purpose, with a reason students can see.

All Google AI products in one place: ai.google/products
The room's take

What you said about the instructor's role

Your words, gathered live. The bigger the word, the more of you said it.

The role
Responses appear here once the room has typed them in.
The examples
Responses appear here once the room has typed them in.
More commonAAAless common
What just happened

That podcast was strategy one

You did not just hear about a technique. You felt it. You listened on purpose because the work was hidden in the audio. You wrote answers about your own course. AI could not do that for you, because the prompts were tied to something only you know: your class.

The throughline

Engagement in the age of AI comes from students creating things to show their learning.

Not summarizing. Not describing. Building. A video, a form, a presentation, a podcast, a redesign, a defense. The product is the proof, and the build is the learning.

Strategy one
1

The podcast with hidden assignment nuggets

The move: feed your week's readings into NotebookLM and have the hosts drop three to five little exercises students must catch by listening.

Why it works: turns passive reading into active listening. Works for any subject. Different tool, different mode, same content.

You just experienced this. The podcast you heard is exactly this method, applied to today's topic.

Active listening NotebookLM
Strategy two
2

Research becomes a build: the Google Form flip

Old version
Research an IT career, write up the requirements and the qualifications, and the path to get there. A summary AI can hand them.
Revamped
Still research (AI fine for gathering), still map the path with a self-evaluation of where they are now. Then they create. Build a presentation. Build a Google Form where they flip roles and become the employer writing interview questions. Or build a video showcasing themselves walking through it, sharing their screen, presenting their research.

Why it works: writing interview questions forces them inside the role. You cannot fake a good question set without actually knowing the job. The product is the proof.

Constructivism Google Forms
Strategy three
3

Spot the mistakes, then create your own video

Week one: you build a spreadsheet with deliberate errors and record a short screen video walking through it. The video lives only in your course, not on YouTube, not anywhere, so AI cannot summarize it. Students watch and pinpoint where you went wrong in the discussion.

Week two (the heart): students record their own screen-and-voice video fixing those errors. They explain each fix out loud as they show it. That is the throughline in action: they create something only they could make.

Why it works: error analysis is high-order thinking, the private video defeats the AI shortcut, and the student video is "teach it to prove you know it."

Error analysis + creation Screen recorder + spreadsheet

Your "spot the mistakes" sample video goes here

Auto-loaded from /videos/professor-errors.mp4. Use the button above to swap.

Strategy four
4

Turn images into classroom video with Google Flow

The move: generate strong images, then turn them into short videos to set up a lesson, illustrate a hard concept, or spark discussion. Custom visuals beat generic stock, and motion holds attention.

Why it works: a 20-second clip you made for this exact lesson lands harder than bullet points, and students remember images far longer than text.

Dual coding Google Flow + Gemini
Step 1 · Generate the image
A red blood cell generated as a still image with Gemini

Start with a still image. I asked Gemini to generate a red blood cell. Took about 20 seconds.

Step 2 · Turn it into video

Then I dropped that image into Google Flow. Hit play. Same red blood cell, now moving.

Strategy five
5

Build your own resources in Claude (this page is exhibit A)

The move: instead of sending students to the open web, build the resource yourself, at college level, right here. A grades dashboard, a practice gradebook with nicknames for privacy, model papers, rubrics, study guides.

Why it works: the resource fits your course instead of a generic one, and students get one trustworthy source instead of ten sketchy tabs.

Claude
The pattern across all five: students still do the thinking, but the work now asks them to build, defend, or judge. Tie it to something personal, local, or live, and AI cannot do it for them.
The room's take

What you said about the student's role

Your words, gathered live. The bigger the word, the more of you said it.

The role
Responses appear here once the room has typed them in.
The examples
Responses appear here once the room has typed them in.
More commonAAAless common
Part three

Teaching students to use AI responsibly

Responsible use is a skill we teach, not a rule we post. Three habits do most of the work.

📝

1. Acknowledge it

Every submission carries a short AI-use statement: what tool, what for. Normalizes honesty instead of teaching students to hide.

🔍

2. Hold it accountable

Students state what they verified, what AI got wrong, what they changed. AI is a source to fact-check, not an authority to trust.

🧠

3. Own the thinking

Judgment, voice, and final calls stay human. AI can draft. The student decides.

Steal this: the AI acknowledgment statement

Put one line like this at the bottom of any assignment where AI is allowed. It takes the secrecy out of the room.

AI Use Statement: I used [tool] to [task]. I checked its output by [how I verified]. I changed [what] because [why]. The ideas and final decisions are my own.

Responsible-use checklist

Click each habit you expect from a student. Watch the ring fill.

States which AI tool was used and for what
Fact-checks every claim against a real source
Flags what AI got wrong or oversimplified
Rewrites in their own voice, not copy-paste
Can explain the work out loud without notes
Knows which assignments are AI-free, and why
0/6
habits expected
The room's take

What you said about the role of AI

Your words, gathered live. The bigger the word, the more of you said it.

The role
Responses appear here once the room has typed them in.
The examples
Responses appear here once the room has typed them in.
More commonAAAless common
Part four

The toolkit: what to use and what for

An honest shelf. Match the tool to the verb, pick two, and go deep.

All Google AI products in one place: ai.google/products
How to choose: match the tool to the verb. Build with Claude. Explain by audio with NotebookLM. Show by video with Google Flow. Read the room with Mentimeter. Everything else is a bonus once those four feel natural.
Live showcase

What would you build for your classroom?

Throw out an interactive component you'd want for your students. We'll pick one and build it together in Claude, live.

A few ideas to spark yours

A visual for a tough concept

Anything you find yourself re-explaining every semester.

A practice quiz that adapts

Wrong answers branch to targeted hints.

A simulation or sandbox

For students to experiment with a process.

A study guide that learns

Pulls together notes, builds quick-reference cards.

I'll take two or three from the room, spin the wheel, and we'll build the winners live in Claude.

Enter your idea here

Be as detailed as possible. What interactive component would you want to build for your classroom? A visualization for a tough concept, a simulation, a quiz, a dashboard, a study tool.

0 ideas submitted ⏱ About 2 minutes Tip: hit Ctrl+Enter to submit faster.
Live: 0 ideas saved from the room
Your ideas appear here. Submit a few, then spin the wheel.
Vote again

Same question. Has anything moved?

Tap your honest reaction now. Compare to the start.

Your honest gut reaction to students using AI?

Tap an option to cast a vote.

🚫 Ban it0
🙈 Ignore it0
🎉 Embrace it0
Total votes: 0

How did the room move?

🚫 Ban it
Before 0% → After 0%
🙈 Ignore it
Before 0% → After 0%
🎉 Embrace it
Before 0% → After 0%
The one ask

Do not solve policy today. Do one thing.

Before next term, pick one assignment and turn it into something students create.

A video, a form, a podcast, a defense, a build. One assignment. Decide its AI rule on purpose.

This whole page was built with AI. So can yours.

1 / 1