Journey through AI: Weekly Lessons from the Undergraduate Classroom

Stepping into the Arena

Dec 10, 2025

This fall I launched something new at George Mason University: UNIV 182 – AI4All: Understanding & Building Artificial Intelligence, the first campus-wide course in AI literacy, open to every undergraduate, regardless of major. It satisfies the Mason Core requirement in Information Technology & Computing, and, more importantly, it’s meant to lower the barrier of entry into AI for every student on campus. This is not an appreciation course. We understand, we apply, we critique, we build. This course has a rhythm. Join us!

This morning marked our final class, our final exam. But it also marked something far more significant: the moment when my students moved from building in protected space to presenting in public space.

When Protected Space Meets Public Scrutiny

I don’t believe in traditional final exams for a course like this. If you can’t defend your work to people who evaluate ideas for a living, you haven’t learned what I’m teaching.

So I designed Final Exam Day as a showcase: structured, high-stakes, public. I invited leaders from industry, nonprofits, and education. These are people I trust, people who ask hard questions, people who understand what it means to build responsibly. They weren’t there to grade. They were there to engage as partners in thought, to ask the question undergraduate students almost never face: Why this? Why now? Why you?

The room setup was deliberate. Six round tables, each one with a student team, at least two guests per table. I started them with warm conversation. Let students find their footing. Then we shifted: ten-minute presentations, strictly timed. Five minutes of structured cross-examination. No hiding behind slides.

After formal presentations, I opened the room for browsing. Guests drifted from table to table, diving deeper into ethics, feasibility, data pipelines, privacy safeguards. The questions got sharper. The students had to think some more on their feet.

And yes, we had pizza. Community matters. Celebration matters.

Six Teams, Six Ways of Seeing the World Through AI

I’m not sharing project details publicly. Several guests told me these teams could credibly move into startup or nonprofit space, and I take my responsibility to protect student work seriously.

What I will tell you: Students could tackle any societal problem amenable to an AI-based solution, but they had to justify everything: the data sources, the model choice, the ethical safeguards, the system design, the feasibility. No hand-waving. No “AI will figure it out.”

The projects that emerged spanned energy efficiency, retail innovation, combating misinformation in advertising, behavioral intervention, transportation safety, and more. But more than the domains, what struck me was the quality of reasoning. These weren’t PowerPoint dreams. These were engineered proposals with risk mitigation, failure modes mapped, and ethical trade-offs acknowledged.

That’s what I’ve been building toward.

What I Saw Today

I’ve been in the room with these students for fifteen weeks. I’ve watched them struggle through checkpoints, redesign after feedback, face conceptual stressors that later became breakthroughs. I’ve watched them learn to think like builders.

But today was different. Today, the room was filled with people who evaluate ideas professionally, and I got to watch my students meet them as equals.

Here’s what happened:

But first, let me tell you who was in that room. I asked each student to introduce themselves with their year and major. Two computer science students. One IT major. Two cybersecurity students. The rest? Economics, finance, business, policy, education, health, kinesiology, etc. This is exactly why I designed the course this way, to prove that AI literacy doesn’t require a STEM background. My guests kept coming back to this. One CTO pulled me aside and said, ‘The kinesiology major just defended a data pipeline better than most engineers I interview.’ Another said: “It is great that you lead with the problem and then the method.” That’s the point.

Students who started the semester saying “I used ChatGPT to…” now speak fluently about training data bias, model selection constraints, risk surfaces, deployment contexts, and mitigation strategies. They don’t talk about AI as magic anymore. They talk about it as engineering, design, and responsibility.

One guest, a nonprofit director, pulled me aside: “These students aren’t just building prototypes. They’re reasoning about impact at a level I don’t see in graduate programs.”

That sentence is why I designed this course.

Another guest, a startup founder: “I’d hire half this room tomorrow.”

I built UNIV 182 around a principle I call AI literacy as agency. Not knowing what AI is. Knowing what it asks of you when you use it in the world. Knowing when to say no. Knowing what questions to ask before you say yes.

Today, I watched that principle become visible.

The Real Work Happened Between the Lectures

I structured this course as a studio, not a lecture hall. Thinking visible. Feedback constant. Misconceptions intercepted in real time.

I taught my students to revise problem statements when their data logic failed. To abandon model architectures that couldn’t solve the task. To surface ethical contradictions in their own designs before someone else pointed them out. To ask, at every stage: , First, would we want it? Would this work in the world? Should it?

Every checkpoint was designed to create productive friction. Every debate was scaffolded to surface conflict. Every revision cycle was timed to force prioritization under constraint.

The syllabus and assignment maps document not just what we covered, but why, the pedagogical rationale behind every choice. I didn’t want a course that works because of my expertise. I wanted a course that works because the structure does the teaching.

When final projects reach this level of coherence, it’s because the work happened in the spaces I designed for it: at the tables during studio time, in the debates I staged, during checkpoint sessions where I pushed until the thinking clarified, in the moments when a student finally asked the right question.

That’s where I teach. That’s where learning lives.

On Endings and Beginnings

Every semester with authentic building ends the same way: you watch a community form around shared intellectual effort, you see students transform into builders. And then the semester ends. Abruptly. Completely.

Today was a celebration. It was also a closing. I feel this every time, and it never gets easier.

But here’s what I know: UNIV 182 is, to my knowledge, one of the first campus-wide AI literacy courses in the country designed for all undergraduates, and I mean all. Economics majors. Education majors. Health and kinesiology students. Only five of my students had any STEM background, and yet every single one of them built, defended, and reasoned about AI systems at a level that impressed industry professionals.

I was so proud of my students today.

As the celebration wound down, several students had to leave for other finals. But those who remained wanted a class photo with our guests. A way to mark the moment, to say: we did this together.

Here they are. I’ve been writing about them all semester. You’ve been on this journey with us. It is good that you see them.

Amarda Shehu

Discussion about this post

Ready for more?