The rapid adoption of generative artificial intelligence (AI) in higher education has prompted educators to reconsider how professional learning and assessment are designed and justified. While AI can improve feedback quality and efficiency, it also raised concerns about academic integrity and learner agency, i.e., the transition of students from passive recipients of information to active participants in their own learning. Traditional assessment practices are increasingly misaligned with contemporary digital competencies. This study examines AI integration in engineering education, with a focus on assessment design and educator reasoning. Using a qualitative approach, the study analysed open-ended survey responses from engineering educators at an Australian university. Participants reflected on their use of large language models for teaching, assessment and professional judgement. The results show that educators value AI for idea generation, but remain concerned about over-reliance, reduced learner accountability, and misalignment between AI use and intended learning outcomes. Educators emphasised the need for deliberately designed assessments that require learners to evaluate AI outputs, justify their decisions using evidence and disciplinary reasoning. The findings show that the educational value of AI depends on pedagogical intentions rather than the technology itself. When aligned with frameworks such as CDIO (Conceive-Design-Implement-Operate), AI-supported assessment can shift the focus from simple task completion toward the development of professional judgement and ethical awareness, positioning AI as a supplement for, rather than a replacement for, professional thinking.
