Learning without limits : analysing the usage of generative AI in a summative assessment

Tools

Clift, Lee and Petrovska, Olga; Kallia, Maria and Crosby, Ryan, eds. (2025) Learning without limits : analysing the usage of generative AI in a summative assessment. In: CEP '25: Proceedings of the 9th Conference on Computing Education Practice. ACM Press, pp. 5-8. ISBN 979-8-4007-1172-5 (https://doi.org/10.1145/3702212.3702214)

[thumbnail of Clift-and-Petrovska-2024-Learning-without-limits-analysing-the-usage-of-generative-AI-in-a-summative-assessment]
Preview
 Text. Filename: Clift-and-Petrovska-2024-Learning-without-limits-analysing-the-usage-of-generative-AI-in-a-summative-assessment.pdf
Final Published Version
License: Creative Commons Attribution 4.0 logo
Download (853kB)| Preview

Abstract

This paper explores how Generative AI (GenAI) can be introduced within summative assessment components in software engineering education. We present an example of an assessment which allows learners to use GenAI in a freeform, constructionist manner, as part of a large, software development project. This work is inspired by previously executed AI-focused assessments and surveys, which explicitly indicate that learners on an Applied Software Engineering Degree Apprenticeship Programme want to formally learn how to use GenAI tools when programming and their employers want to see these skills from graduates. The learning outcome of the assignment was for learners to explore a typical developmental pipeline as a solo developer, moving from design to development to finished product. Learners were marked exclusively on their end product and understanding of application components, not the written code itself, resulting in an assessment where the end product and project were prioritised over foundational code (which was adequately assessed in other components). The results show that all learners used GenAI to some extent during their project, and in all cases, they found it beneficial for large programming tasks. Learners were generally able to produce a larger, more comprehensive and more ambitious project, compared to previous years. It is proposed that removing the barrier to GenAI - and demystifying it - can encourage a constructionist approach to its use, and normalise it as a potential tool for programming.

ORCID iDs

Clift, Lee ORCID logoORCID: https://orcid.org/0000-0002-8313-9934 and Petrovska, Olga; Kallia, Maria and Crosby, Ryan
CORE (COnnecting REpositories)