Date
Publisher
arXiv
As Generative AI (GenAI) disrupts higher education, institutions increasingly require students to declare AI use. However, generic, binary declarations (e.g., "I used GenAI") fail to capture the nuanced application of these tools in different academic tasks. Establishing transparency is key to protecting academic integrity, promoting AI literacy, and shifting the focus from policing to professional practice. In response, this paper contributes a design artefact and an accompanying position: a framework of two task-specific declaration structures, one for writing-focused activities and one for coding assessments, developed for a Computer Science department on the basis of an existing taxonomy of GenAI usage, together with an argument that task-specific disclosure is needed to move beyond binary declarations. By categorising AI usage across specific cognitive and developmental stages, such as structural planning vs. Textual Content Generation, or code improvement vs. code generation, the framework encourages students to reflect on their own learning process and clarifies the boundary between acceptable assistance and academic misconduct. We propose this domain-specific approach as a foundation for fostering more honest assessment in Computer Science and other disciplines, aiming to better prepare students for professional environments where documenting GenAI workflows might be an essential job requirement.
Study at a glance
- Question: How can domain-specific declaration frameworks structure transparent student reporting of generative AI use in writing and coding assessments?
- Setting and data: Computer Science department at Swansea University, Wales.
- Method: Design artefact and conceptual framework development translating an existing taxonomy of generative AI usage into task-specific assessment tools.
Why it matters
Generic, binary AI declarations fail to capture the nuanced ways students use generative AI. This prevents declarations from serving as integrity safeguards, reflective prompts, or useful feedback for educators.
Key findings
- A writing-focused declaration structure captures student generative AI engagement across six categories: planning and structure, textual content generation, research and analysis, revision and polish, visual content, and evaluation and feedback.
- A parallel coding-focused structure organises generative AI usage across five software development categories: planning and design, code generation, code improvement, understanding and learning, and documentation and reporting.
- Both declaration structures capture the extent of generative AI assistance across a three-point scale alongside prompt-level explanations to distinguish acceptable learning support from academic misconduct.
Summary drafted with AI from the full paper and may contain errors. See the full paper for details.
What is the application?
Who age?
Why use AI?
Study design
