The AI Teaching Value Alignment Canvas: A Lightweight Framework and Rubrics for Connecting AI-enhanced Pedagogic Methods and Program Target

AI in Education is a rapidly evolving field and program teams often do not have a shared language to discuss the values, evidence, safeguards and scalability of individual learning activities (ILAs) incorporating AI. This study explored the feasibility of the AI Teaching Value Alignment Canvas (ATVAC) which is a one-page instrument that serves as a tool and a rubric to decide whether ILAs integrating AI are suitable for implementation at the program level. The Canvas was developed using design-based instrument development where the authors reviewed the course materials, refined and revised the Canvas based on the program learning outcomes, assessment rubrics and quality assurance materials. An expert panel of faculty members participated in a micro-Delphi method to validate the content and usability of the Canvas. A multi-rater scoring study was conducted in a pilot test to investigate the validity and reliability of using the Canvas to score 12 cases of AI-based teaching practices and found that the percent agreement, weighted Kappa and total-score ICC of the items on the Canvas were higher for more concrete and anchored items (e.g., Problem clarity: κ=0.78, Safeguards strength: κ=0.76, Improvement loop: κ=0.73) and lower for more interpretive items (e.g., Indicator validity: κ=0.41, Scalability & cost: κ=0.39). The total-score ICC of the Canvas is 0.79. The Canvas addresses four values which are the learning outcomes, the teacher workload and decision quality, the service experience, and the program quality and consistency. It has feasible and practical evidence indicators which do not require the recording of fine-grained details of the ILAs integrating AI. The Canvas also has clear risk safeguards and governance arrangements that are lightweight. The Canvas will support program teams in making consistent decisions on the adoption of ILAs integrating AI and thus implementing AI-based pedagogies in a responsible manner across a variety of teaching contexts.