Amidst the normal fanfare at Stanford’s 136th Convocation, Vice Provost for Undergraduate Education James Hamilton made an unusual request during his remarks. Hamilton recommended students enroll in a new course, CS193T: “Thinking With AI,” so they could learn “agentic workflows.”
Along with CS193V: “Effective Vibecoding,” CS193T is part of the computer science (CS) department’s push to ensure students with non-technical backgrounds can engage with large language models (LLMs). According to the class’s website, CS193T aims to teach “advanced prompting skills,” where students are instructed to apply AI to the work they normally do, but with “layers of reflection, analysis and comparison.” CS193V teaches students without coding experience on how to “vibecode” — the new art of getting an LLM to code for you, without review of the code itself.
Keith Schwarz, who leads teaching in CS193V, said that Chris Gregg, associate professor of CS and the department’s associate chair for education, approached him last spring about teaching an AI-assisted coding class “aimed at a general audience.”
CS193T has no prerequisites, unlike most courses the department offers. CS193V’s only prerequisite is CS106A or an equivalent level of a few months of programming experience. Even so, the syllabus starts from the basics.
“The vision of the course I’m trying out this quarter doesn’t involve reading or writing code,” Schwartz said. “Having a basic knowledge of coding is nonetheless helpful for understanding certain concepts we’ll cover.”
Amari Porter ’27, a teaching assistant (TA) for CS193T, said that the class attracted a wide range of students, from “incoming frosh who might not know much, to mature Distinguished Career Institute fellows, who also might not know much.”
She said that CS senior lecturer Cynthia Bailey and the TAs are developing assignments as the class progresses. According to Porter, the assignments included “looking at differences between the major three [AI] models and having models infer life details when given writing samples.”
Bailey had previously worked in AI safety and was chosen by the CS department because she is “very big on thinking about the impacts of AI usage,” as Porter noted. TAs, too, came from a range of experiences with AI as well.
“When I was a frosh, ChatGPT had just started to get popular, and during sophomore and junior year, I had felt overly reliant on AI and questioned how much of my own agency was being automated away,” Porter said, adding that she “tries to teach how to use AI without relying on it like a crutch.”
Haowen Chen ’28, an environmental engineering major, decided to take CS193V after seeing projects his friends had vibecoded.
“I also saw a lot of interesting things people were building on Instagram,” Chen said. “I have several projects myself, but I felt like I was lacking some of the basic technology and pipelines.”
Schwarz told The Daily that vibecoding is “absolutely” a teachable skill. Even if students weren’t going to write the code themselves, he said, students would still want to have a basic understanding of the technology in order to evaluate costs and maintainability and determine if software works correctly.
“I’m hoping CS193V will equip students to think critically about these questions and make appropriate decisions,” Schwarz said.