The AI Assessment Scale for Classrooms

The AIAS was first published in 2023 and piloted in higher education. That version had five levels: No AI; AI-Assisted Idea Generation and Structuring; AI-Assisted Editing; AI Task Completion, Human Evaluation; and Full AI. In 2024, the authors released an updated version with simpler names: 1. No AI: work done without AI help. 2. AI Planning: AI for brainstorming, research and structuring, but not the final work. 3. AI Collaboration: AI helps with drafting and feedback, and students critically evaluate and change its output. 4. Full AI: AI may be used throughout, with students directing it. 5. AI Exploration: creative or new uses of AI as part of the task itself. The update made two points clear. First, the levels are not a ladder of quality. A Level 4 task is not easier or less rigorous than a Level 1 task. It assesses different skills. Second, a label cannot guarantee anything by itself. If students do an assignment at home, nobody can confirm that "No AI" was followed. Level 1 only means something when the work is done under supervision. That second point links the scale to wider thinking about assessment security. The University of Sydney's "two-lane" approach, for example, separates secure, supervised assessments from open ones where AI use is expected and taught. To tell students which level applies, teachers usually put it on the assignment brief, explain why that level fits the learning goal, show examples of acceptable and unacceptable use, and require a short disclosure at any level above 1. Common misconceptions are that a whole course must sit at one level, and that the scale is a detection or enforcement tool. It is a communication and design tool.