Building Ethical AI Standards for Education

In this workshop, we will explore the technical and pedagogical foundations necessary for AI platforms to deliver responsible quantitative assessments. We will start by building an understanding of how large language models (LLMs) operate, focusing on the concept of token prediction. This foundation will lead us to examine ways to evaluate how confident an AI model is when giving a response. This will prove key insights into their reliability and limitations. Next, we will shift to a hands-on exploration where participants will experiment with different prompting strategies and AI models. This interactive segment will allow us to observe firsthand how these variations impact the accuracy and consistency of grading assessments. Finally, we will conclude with a collaborative discussion on the crucial role educators play in shaping guidelines for responsible AI use in assessment. We will explore how educators can—and must—take an active role in ensuring AI assessments are fair, ethical, and pedagogically sound.