Harvard embraces generative AI in the classroom, adopting it as an official learning tool for its flagship coding course.
Starting this fall, students enrolled in Computer Science 50: Introduction to Computer Science (CS50) will be encouraged to use AI to help them debug code, give feedback on their designs, and answer individual questions about error messages and unfamiliar lines of code.
“Our own hope is that, through AI, we can eventually approximate a 1:1 teacher [to] student ratio for every student in CS50, as by providing them with software-based tools that, 24/7, can support their learning at a pace and in a style that works best for them individually,” says CS50 professor David J. Malan, as reported by The Harvard Crimson.
It's a swift turnaround from the last school year: Harvard did not have an AI policy at the end of the fall 2022 semester.
The new approach will not use ChatGPT or GitHub Copilot, both of which are popular among programmers. Malan says the tools are "currently too helpful." Instead, Harvard has developed its own large language model, a "CS50 bot" that will be "similar in spirit," but will focus on "leading students toward an answer rather than handing it to them,” he says.
CS50 is also available for non-Harvard students to take on the online platform edX. The new AI policy will extend to the edX version. "Even if you are not a student at Harvard, you are welcome to “take” this course for free by working your way through the course’s eleven weeks of material," says the site. Teachers at other institutions can also license the material for their own courses.
“Providing support that’s tailored to students’ specific questions has long been a challenge at scale via edX and OpenCourseWare more generally, with so many students online, so these features will benefit students both on campus and off,” Malan says.
The adoption of generative AI by one of the nation's top universities adds a new wrinkle to the use of tools like ChatGPT in an academic setting. Since ChatGPT launched in November 2022, teachers and professors have struggled with students turning in work generated by it.
One professor at Texas A&M University-Commerce refused to grade work he believed was "ChatGPT shit." Tools to detect AI-written work have also gained popularity, but they focus on text, not programming languages.
“We’ll make clear to students that they should always think critically when taking in information as input, be it from humans or software,” Malan says. “But the tools will only get better through feedback from students and teachers alike. So they, too, will be very much part of the process.”