Brown University launched an experimental Agentic Studio course this spring to teach computer science students how to work with AI coding agents. The initiative addresses a growing industry demand for professionals who can manage autonomous coding tools while mitigating the risks of flawed AI-generated software in academic research.
Kathi Fisler, research professor and director of undergraduate studies for Brown's Department of Computer Science, anticipated this shift. She recognized that agentic AI tools improved suddenly, making them viable for real tasks. As students enter the job market, they will be expected to know how to use these tools, making AI for Education a critical focus for academic curriculum design.
"We'd been seeing the whole agentic AI coding thing coming for a couple of years," Fisler said. "It was there, but it wasn't yet really good at what it does. Then suddenly the capabilities of the tools jumped, and we knew we had to teach the students how to work with these things."
Managing the risks of flawed output
The tools are now capable enough to cause real harm if used incorrectly. AI-generated code remains far from flawless. At Brown, students frequently write code for research projects outside the computer science department or for community nonprofit organizations.
"If students with limited programming experience are going to use agents when they work with professors on research projects or off-campus partners, now there's an opportunity to do real damage," Fisler said.
A controlled classroom experiment
Fisler worked with computer science professors Shriram Krishnamurthi and Michael Littman to incorporate agentic programming into the curriculum. Instead of overhauling all introductory classes, the team created a measured, project-based approach. The new class gave students with prior computer science coursework a supervised environment to evaluate Generative Code tools, learning where they add value and where they fail.
"We were upfront with the students about the fact that this was experimental," Littman said. "We said: 'We're going to try to figure out how to teach this stuff, and you're going to be our partners in exploring this space.'"
Students worked in teams on practical software tasks, such as creating a graduation requirement checker or developing a messaging app. Throughout the semester, they kept journals documenting their specific experiences using the AI agents on each project.
Why this matters for educators
Academic departments cannot ignore AI coding assistants without leaving students unprepared for modern software roles. However, allowing unsupervised use introduces severe risks to external research partners and community organizations. Educators must build structured, supervised modules that teach students to audit and correct AI output, rather than banning the technology outright.
Your membership also unlocks:
