This is an interesting short post from colleagues at the University of Sydney considering the challenge of redesigning assessments due to the rise of generative AI in software. It argues against the possibility of making “AI-proof” exams. Instead, it suggests a dual strategy that includes secure environments for some assessments and adapts others to incorporate AI tailored by discipline. It advocates this shift to focus on the human aspects of education and teaching students to engage with AI-generated content critically. The article advises designing relevant tasks and oral assessments as AI evolves, ensuring that assessments are future-proof.
What to Do About Assessments If We Canāt Out-Design or Out-Run AI
Latest posts by Gunter Saunders (see all)
- Presentation on AI and the Future of Higher Education - 17 May 2024
- Experimental evidence on the productivity effects of generative artificial intelligence - 25 March 2024
- What to do about assessments? - 19 February 2024