Should assessment change when students have access to generative AI?
The widespread availability of generative AI tools such as ChatGPT is changing how students learn, research, and complete academic work.
As a result, traditional assessment methods need to evolve rather than simply trying to prevent students from using AI.
First, assessments should focus more on critical thinking, application, and problem-solving rather than memorisation. AI can quickly generate definitions, summaries, and standard answers, so questions should require students to analyse information, justify decisions, solve unfamiliar problems, or apply concepts to real-world situations.
Second, educators should consider process-based assessment. Instead of evaluating only a final submission, teachers can assess drafts, research notes, reflections, presentations, and discussions. This provides insight into how students developed their ideas and makes learning more important than simply producing a polished final answer.
Oral assessments and classroom discussions can also become more valuable. Asking students to explain, defend, or critique their work allows educators to evaluate whether they genuinely understand the subject. Short viva-style assessments can complement written assignments.
At the same time, students should not necessarily be prohibited from using AI. Instead, assessments can teach responsible and transparent AI use. Students could be asked to use AI for brainstorming or feedback and then identify its limitations, verify its information, and explain how they improved the AI-generated output. This develops AI literacy, which is increasingly important in higher education and the workplace.
Finally, assessment should become more authentic and personalised. Projects connected to students’ experiences, local communities, current issues, or practical workplace situations are harder to complete through generic AI-generated responses and provide stronger evidence of individual understanding.
In conclusion, generative AI does not make assessment less important-it makes better assessment necessary. The goal should shift from detecting whether students used AI to evaluating what students understand, how they think, and how responsibly they use technology. Assessments should therefore measure reasoning, creativity, communication, application, and the learning process alongside knowledge.
The writer is a Campus Director and Professor, NMIMS; Views presented are personal.
