ARAD 2.0 (AI-Responsive Assignment Design for Human Learning) asks one question: does your assignment still measure what it claims to measure when a student uses AI? This workflow arrived in update 0.8.1. Get the update if you do not see it.
Steps
- Select Design with ARAD 2.0 in the task list.
- Enter the course and learner level, then the Goal: what students must be able to do.
- Optional: paste the current assignment, or import it with Use my document.
- Choose the AI mode (AI-Restricted, AI-Limited, AI-Collaborative, or AI-Required) and an architecture pattern (AI Sandwich, Verification Challenge, Comparative Analysis, or Generative Challenge). You can also ask for a recommendation.
- Select Build ARAD design.
What you get
The draft follows the GOALS steps: Goal, Openness (the mode with permitted and prohibited AI uses), Architecture (formative and summative phases), Look for Evidence (a rubric and a Transfer Gap Check), and Sustain (an AI literacy lesson, a pilot plan, and a faculty pre-test).
Before you use it
- Read the design closely. The local model can make mistakes, especially in rubric wording.
- Run the faculty pre-test: give the assignment to current AI tools and score their output with your rubric. If AI output alone reaches Proficient, redesign.
- Design an assignment also offers the four ARAD modes when you want a quicker draft.
Learn more about the framework in the ARAD 2.0 article.
Last updated October 7, 2026