Seminar: Supporting Novice Programmers in Transitioning into the Age of Human-GenAI Collaborative Coding

Abstract

The rise of Generative AI (GenAI) is transforming computing education. In the past, introductory computing courses were primarily concerned with equipping students with the skills needed to write code independently. However, in the age of GenAI, many are suggesting that skills such as the ability to describe computational tasks through natural language and comprehend the resulting generated code should receive further emphasis.

This talk explores David’s work on developing autograding and feedback mechanisms for Explain in Plain English (EiPE) questions. These activities are aimed at developing code comprehension, providing students a segment of code and asking them to describe, in natural language, that code’s purpose. In particular, it will focus on David’s ongoing work on developing and evaluating approaches for leveraging large language models to autograde and provide actionable feedback for these questions at scale. By utilizing modern large language models, EiPE questions have been transformed into Explain in Plain Language (EiPL) questions, broadening their availability to a wider range of natural languages. In particular, this talk will touch on David’s ongoing work in investigating effectiveness of LLM based autograders for EiPL questions in the context of Indic languages.

_Biography

David Smith is a 5th-year Ph.D candidate at the University of Illinois Urbana-Champaign in the Siebel School of Computing and Data Science. His research focuses on developing and assessing innovative tools and techniques to support novice programmers. His current work explores topics such as the integration of generative AI tools in programming education, the development and evaluation of multilingual code comprehension activities, and Parsons problems—a programming activity designed to scaffold the process of learning to write code.