Why it works
Generative AI can produce competent responses to many traditional take-home assignments, which means some of them no longer reliably reveal what a student knows. Rather than only policing this, emerging practice suggests rethinking assessment: designing tasks that make student thinking and process visible (in-class work, drafts, discussion, explanation, application to novel or personal contexts), or deliberately incorporating AI with the assessment focused on the student's critique, judgment, and reasoning. The goal is assessment that still measures genuine learning in a world where AI exists.
The research: Emerging practice on assessment and generative AI.
The run-of-show
Choose your slot. The agenda, timings, and length update to match.
The core activity: redesign an assessment for an AI world
Teachers leave with one assessment redesigned to reveal genuine thinking.
- Pick an assessment AI could now complete.
- Redesign it to reveal genuine thinking: make process visible (drafts, in-class, explanation), use novel or personal contexts, or build AI in with the focus on student judgment.
- Make sure it measures real understanding.
- Plan how you will know it reveals the student's own learning.
Facilitator notes
- Recognize what AI can complete.
- Make thinking and process visible, or use novel and personal contexts.
- Or build AI in, focused on judgment.
Adapt it
- Elementary: process, explanation, in-class evidence.
- Secondary: process, novel contexts, AI-integrated with judgment.
- All settings: reveal genuine thinking.
Participant handout
One page for every teacher. Print it, or save it as a PDF.
Assessment in an AI World: assess the thinking
Reveal real learning: make process visible, use novel contexts, or build AI in focused on judgment.
- The assessment AI could now complete:
- How I redesign it to reveal thinking:
- The process or context that makes it genuine:
- How I might build AI in focused on judgment:
- How I ensure it measures real understanding:
Make it stick
Assess genuine thinking:
- Assess genuine thinking.
- Redesign AI-vulnerable tasks.
- Bring a redesigned assessment to the next PLC.