AI-enabled assessment: framework, perspective and educational outcomes

The development and use of artificial intelligence (AI) tools such as ChatGPT have raised concerns about assessment practice in higher education (HE). Prior studies have discussed the need to re-imagine and re-design assessment instruments; however, there is limited empirical evidence on how to effectively deploy this in practice. This paper proposes the TSD (task, scaffolding support and deliverable) framework to guide academics and education providers on how to effectively integrate AI and reflective practice into assessment design. By doing so, we adopt a mixed-method approach to investigate educational outcomes, tutors’ and students’ perspectives on the reflective AI-enabled assessment. This study collected data from UK University students (22 respondents) and tutors (24 participants) using a survey questionnaire and semi-structured interview questions, respectively. This study provides empirical support for positioning reflective practice as evidence of learning rather than being treated as a mere assessment activity. The findings show that AI output verification specific tasks need to be incorporated into AI-enabled assessment design. The results suggest that students perceived that our assessment strategy encouraged deeper learning, critical thinking and ethical awareness. Similarly, tutors were mostly positive with the adoption of the strategy, but some highlight the need for caution and multimodal approach .