This chapter investigates the use of ChatGPT for designing algebra exams in higher education, focusing on how educators’ mathematical knowledge shapes AI-generated assessments. Using Bloom’s Taxonomy and the Mathematical Knowledge for Teaching (MKT) framework, the study analyzes interactions between three university educators and ChatGPT, evaluating the cognitive alignment and pedagogical quality of the generated tasks. Findings reveal that ChatGPT effectively produces procedural and analytical tasks but faces challenges in creating higher-order thinking tasks, such as those requiring synthesis and creativity. Educators’ mathematical knowledge for teaching was critical in refining AI-generated content, with iterative feedback leading to more cognitively aligned and pedagogically sound exams. The study underscores the importance of integrating AI capabilities with educator knowledge to design effective assessments and highlights the need for structured frameworks to guide AI integration in higher education assessment.