Harvard University’s flagship computer science course, Computer Science 50: Introduction to Computer Science, has embraced the use of artificial intelligence (AI) to enhance the learning experience for students. The course, which currently has around 70 students in its summer program and expects over 600 in the fall, is utilizing ChatGPT, developed by OpenAI, to free up teaching assistants (TAs) and allow them to focus on providing personalized support to students.
David Malan, the Gordon McKay Professor of the Practice of Computer Science at Harvard, expressed the university’s enthusiasm for integrating AI into their curriculum. He emphasized the need for additional guidance in computer science classes and explained that the aim is not to replace teachers but to enhance their effectiveness by reallocating resources. By leveraging ChatGPT, Harvard is able to assist students in understanding lines of code, improving coding style, and answering frequently asked questions.
The adoption of AI in education is seen as a positive change by Lynn Parker, director of the AI Tennessee Initiative. Parker believes that using AI can help address the shortage of TAs in computer science and enable more students to access personalized assistance. She sees the goal of providing easy access to support for every student as a fantastic objective.
Harvard is among a growing number of institutions embracing AI in various capacities. Admissions offices, such as Georgia Tech, are experimenting with AI, while other institutions have introduced AI faculty workshops. Some universities, including Purdue University, Emory University, and the University at Albany, have even hired faculty members dedicated solely to artificial intelligence.
Malan envisions expanding the use of AI beyond introductory computer science at Harvard. He plans to introduce additional applications, including AI explanations for error messages, bug detection in code, program design assessment, and evaluating students’ understanding through AI conversations. These features aim to guide students in the right direction by asking rhetorical questions, similar to how TAs would assist them.
Parker believes that the Harvard course could set a standard for wider adoption of AI in education, particularly among tech-savvy students. However, Fiona Hollands, founder of the research and evaluation organization EdResearcher, acknowledges that some faculty members may be resistant to change. Hollands points out that institutions of higher education often prioritize tradition over efficiency.
Malan highlights the potential of AI in evaluating subjective aspects of coding, which can be time-consuming for TAs. He notes that students typically spend less than 14 seconds reviewing TAs’ comments on their code, making it an inefficient use of time for both parties. By leveraging AI, TAs can have a more significant impact on the student experience.
In addition to freeing up TAs, the availability of AI support aims to help students troubleshoot issues more efficiently, reducing frustration and potentially improving retention rates. However, Malan emphasizes that AI is not infallible and is constantly evolving. Students are reminded not to blindly accept AI answers unless they have been approved by a TA.
While currently limited to computer science courses, Malan hopes to expand the use of AI to other disciplines, including humanities. He aims to demonstrate that AI can have a positive impact across various fields, not just STEM subjects.