Milky Ways and Machine Learning: A Changing Classroom

In the basement of Carroll Hall, the home of UNC-Chapel Hill’s Hussman School of Journalism and Media, Scott Geier enters a windowless classroom. Students sit at desks with big monitors, but each screen is blank — they’re using personal laptops instead.


There are a few minutes left until class starts, and Geier tosses Milky Way bars to students as a reward for arriving early. When the clock strikes 12:20, Geier brings out a red box with golden embroidery. It is about 100 years old, and it belonged to his grandfather. Geier carries the box to the center of the room and sets it on a table. He opens the lid, propping it on the wooden stamp inside the box, and sets the scene.

The stamp depicts two serpentine dragons spiraling around one another, like a double helix. Beside the stamp is a small, circular ink reservoir made of blue and white porcelain.

Geier tells his students to replicate the scene with a generative AI (GenAI) image generator. He puts on the class playlist, which compiles the class’ favorite music, and gives them two songs’ worth of time. Whoever generates the best lookalike, he says, wins a Milky Way. When time is up, he reviews the submissions, focusing on how good prompts make good images. In the end, to the victor goes the spoils (a Milky Way).

Scott Geier, an assistant professor at UNC-Chapel Hill’s Hussman School of Journalism and Media, believes his students will need GenAI competency in all aspects of their lives — both at work and at home. He wants UNC-CH to prepare students for that reality. He not only encourages his students to use GenAI, he assigns them to do so.

When OpenAI released ChatGPT on Nov. 30, 2022, it went viral. At the time, it was the fastest growing consumer application in history. By the end of January 2023, the chatbot had accumulated over 100 million monthly users, according to a UBS study quoted by Reuters in 2023. Building on OpenAI’s work, other tech companies developed their own GenAIs.

Today, many GenAI tools are free and easy to use. With a bit of prompting, they can generate essays, practice tests, computer code and much more. Naturally, students gravitated towards GenAI.

Some educators oppose students’ use of GenAI. Destiny Peterson, a current doctoral student in the Media and Communication program at UNC-CH, is one such educator. She has been teaching “Media Ethics” at the j-school since summer 2023, and she doesn’t allow GenAI in her classroom.

Peterson’s main reason for avoiding GenAI thus far comes down to students’ education.

She wants students to think through problems critically, not simply edit a GenAI’s material. Also, she doesn’t yet know how the media industry will implement GenAI, so she isn’t sure what to teach to prepare her students for a career.

However, Peterson emphasizes that students should be adaptable. They should use GenAI outside of classes that don’t allow it, but avoid relying too heavily on it in classes that do. As for educators, she thinks it all comes down to how they use it. “I support professors doing what they think is best for students,” she says.

Geier thinks what is best for students is learning how to use GenAI.

Scott Geier is a father (of cats and kids), a husband, a musician, a web developer and a supporter of GenAI in the classroom. He is part of the Provost’s AI Acceleration program, which aims to prototype and apply GenAI in research, instruction and operations at UNC-CH, according to its website. Geier earned a bachelor’s degree in philosophy before pursuing a master’s in interactive media at UNC-CH. In 2017, while working toward his master’s, he was offered a position as an adjunct professor. As soon as he started teaching, he got the bug and never looked back. He became a full-time assistant professor in 2023.

Geier says he is not a GenAI expert. He may not know the nuts and bolts of GenAI systems, but because he actively incorporates them into his courses, he knows how they impact the classroom.

Geier has tried out several ways to integrate GenAI into the classroom. He’s tasked students with replicating real objects, storyboarding shots for documentaries and exploring ideas for website designs with GenAI image generation. He encourages students to generate articles, fix broken code and practice simulated interviews with GenAI. In the “AI playground,” students participate in various activities to learn GenAI’s capabilities. One such activity tasks students to solve Wordle using only GenAI. Some things stick, some don’t.

But before any of this is submitted, Geier expects students to edit, factcheck and credit everything that is GenAI generated.

Katrina Angus, a senior studying Advertising and Public Relations, took “Introduction to Interactive Media” with Geier in Spring 2023. She said he was one of her most authentic professors at UNC-CH.

In Geier’s course, Angus learned GenAI’s strengths and weaknesses. Before taking Geier’s course, she used GenAI to write whole essays, which she would then edit. Today she does the opposite.

Geier incorporates GenAI into his courses for two main reasons. First, he thinks it helps students learn better and quicker. Second, he thinks GenAI is changing the world and every industry, and students should be ready for it. “It's not our job to shield our students from reality,” he says.

Both courses Geier teaches, “Introduction to Digital Storytelling” and “Foundations of Interactive Media,”  introduce students to entry-level coding for websites. After ChatGPT was released, he expected students’ work to improve drastically. But it did not. Though his students could ask GenAI for help, they didn’t understand the code itself, which prevented them from fixing mistakes or making edits.

He realized that coding still needed to be taught, but in a new way. Educators needed to teach traditional coding and GenAI-assisted coding side by side. Once he started teaching like this, he did see the drastic improvement he had expected GenAI to bring.

“I've had students who, using AI, learned a ton and built something amazing they would not have been able to build otherwise,” he says.

Geier believes GenAI could change everything. He sees the leaps and bounds GenAI is making and can’t help but imagine a future in which it takes countless jobs from real people. He thinks AI may learn to improve itself, eventually surpassing human intelligence. Such a notion might be written off by some, but Geier thinks it’s a possibility, and that it may happen soon. Though the future is unknown, he wants to prepare his students for whatever lies ahead.

All he can do right now is prepare his students, peers and anyone who will listen. He teaches AI in the classroom, but he always follows one rule: Humans first, GenAI second.