This project will examine how large language models (LLMs) will affect computing education. LLMs and their applications such as GitHub Copilot can create new high quality text based on textual input, but their impact on computing education has not been studied. The project will study the benefits that LLMs can bring to computing classrooms and the potential challenges their introduction can entail focusing on two aspects: studying how students use Copilot and similar tools and exploring the capabilities of LLMs for creating novel course exercises of various types (programming exercises, code explanations, and multiple-choice questions). While Copilot and similar tools help professional programmers write code faster, it is not known whether they are a help or a hindrance for novices who are just learning to program. Being able to automatically generate vast numbers of exercises would greatly alleviate instructor workload, leaving time for other teaching activities.

2023

2027