7. The AI Assessment Scale: Ethical use of GenAI in assessment

Authors:  Mike Perkins, Viet Anh Nguyen Vu, Jason MacVaugh

Institution: British University Vietnam 

Situation

British University Vietnam (BUV), a private higher education institution with approximately 2,500 students, faced an increase in academic misconduct cases related to the misuse of GenAI tools following the release and popularisation of ChatGPT.

Students quickly adopted these tools and integrated them into their writing practices; however, this was done without an understanding of the potential to commit academic integrity violations. This led to a high number of cases in which students received penalties for misusing these tools. BUV needed to address this misuse, while also recognising the inevitable future integration of GenAI into society.

Task

The task of the team leading the integration of AI into BUV was to develop a clear framework explaining how students were expected to use GenAI tools in university assessments to promote ethical and responsible use of AI. Our specific objectives were as follows: 

1. Help educators consider how their assessments might need to be adjusted to account for the integration of GenAI tools into HE.

2. Clarify how and where GenAI tools can be used in student work.

3. Support students in completing assessments in line with the principles of academic integrity, thereby reducing the number of academic misconduct violations within BUV.

Action

To achieve these objectives, we developed and launched the Artificial Intelligence Assessment Scale (AIAS) (Perkins et al., 2023), a five-point scale for using AI in assessments. The AIAS was developed from the work of Furze (2023), who had initially created a scale designed for K-12 use. This was adjusted to account for the specific requirements of BUV as an HEI, leading to the development of the AIAS ranging from Level 1 ("No AI") to Level 5 ("Full AI"). Each level demonstrates a particular allowed usage of GenAI in assessments to provide clear guidelines on how students can incorporate AI tools ethically. The AIAS is shown in Figure 1 below. 

Figure 1. The AI Assessment Scale (Perkins et al., 2024) CC BY-ND 4.0

The AIAS allows us to communicate to students how GenAI may be used in assessments, with academic staff selecting the appropriate level on the scale, depending on the requirements of the course. This approach promoted accessibility and inclusivity by ensuring that all the students had a common understanding of the expectations associated with GenAI use. It also has specific benefits for supporting English as Additional Language (EAL) speakers who make up the majority of the student body at BUV. The AIAS is adaptable and scalable across both disciplines and institutions, and can be modified depending on the specific contexts in which it is employed. This makes it a sustainable solution for any institution facing similar challenges related to GenAI misuse, who wish to integrate AI into learning and assessment.

Results

Following the implementation of the AIAS and related policies at BUV, the percentage of GenAI academic violations decreased from over 100 in the semester prior to the introduction of the scale to zero cases following its implementation. Comparing student outcomes on a semester basis, we saw the pass rate increase by 33.3%, with mean grades increasing by 5.9% on a year-on-year basis (Furze et al., 2024). While the grade increase is modest, the drastically higher passing rate suggests that GenAI helps BUV’s majority of EAL students overcome language barriers and express complex concepts without fear of penalties.

The AIAS also created opportunities for student creativity and engagement by allowing alternative assessment formats, such as presentations including AI-generated visuals used to summarise complex ideas. These options help level the playing field for EAL speakers, who might otherwise struggle to express their depth of understanding in a written task. Similarly, academic staff adopted pedagogical changes to incorporate GenAI into the classroom. Discipline Leaders observed positive shifts in incorporating GenAI tools into modules either as ‘co-pilots’ or for supporting students. The AIAS has inspired adoption among higher education and K-12 institutions worldwide, with several variations of the scale already being tested on a global basis (cf. Devlies, 2024).

Stakeholder Commentary

The AIAS implementation at BUV has received positive feedback from students and faculty, promoting academic integrity and preparing students for the future use of AI in employment. This empowered students to explore multimodal GenAI in their assessment submissions, for example, by enhancing presentations using AI video generation tools and AI-generated characters. Our UK partners have also commended BUV for its proactive approach to addressing GenAI challenges, with some considering implementing the scale themselves. Overall, the AIAS fosters innovation, creativity, and digital literacy while supporting the challenges of academic integrity in a new GenAI environment.

References 

Devlies, E. (2024) ‘AI-gebruik in taken en opdrachten’, Schoolmakers, 29 January. Available at: https://www.schoolmakers.be/blog/ai-gebruik-in-taken-en-opdrachten/ (Accessed: 7/4/24).

Furze, L. (2023) The AI Assessment Scale: From no AI to full AI – Leon Furze. URL https://leonfurze.com/2023/04/29/the-ai-assessment-scale-from-no-ai-to-full-ai/ (accessed 6/6/23).

Furze, L. et al. (2024) ‘The AI Assessment Scale (AIAS) in action: A pilot implementation of GenAI supported assessment’. arXiv. Available at: https://doi.org/10.48550/arXiv.2403.14692.

Perkins, M., Furze, L., Roe, J., MacVaugh, J.(2024). The Artificial Intelligence Assessment Scale (AIAS): A Framework for Ethical Integration of Generative AI in Educational Assessment. Journal of University Teaching and Learning Practice, 21(6). Available at  https://doi.org/10.53761/q3azde36

Author biographies

Mike Perkins

Mike Perkins is the Head of the Centre for Research & Innovation at British University Vietnam. He researches Generative AI's impact on academia, focusing on policy responses and academic integrity.

Jason MacVaugh 

Jason MacVaugh is Dean (Higher Education) at British University Vietnam. He is working with faculty on revising pedagogic approaches in the light of AI tool availability and has worked to change university policy on AI use in student work.

Nguyen Vu Viet Anh 

Nguyen Vu Viet Anh is a third year student at British University Vietnam. He supports faculty in data analysis and research as part of the BUV Centre for Research & Innovation.