An increase in Academic Integrity (AI) breaches has resulted in higher education institutions implementing solutions to improve AI competence. It has been argued that to improve students’ AI understanding, concepts and skills should be taught at the classroom level and contextual factors should be considered. This article presents an investigation on how AI is taught at the classroom level across a range of disciplines, how contextual factors inform approaches to AI education, and how the approaches align with evidence-based recommendations. Purposeful sampling procedures were employed to select units of study from disciplines at two Australian universities. Qualitative data collection methods were used to capture ways AI education was approached and the collected data were analysed through grounded theory methods. The findings show that AI was primarily taught through explicit instruction and personal storytelling and assessed through summative assessment. Such approaches appear to be influenced by personal philosophies, institutional mandates and student backgrounds. While the approaches align with the notion that best practice includes an educative approach, other facets of best practice that have been promoted to combat the rise in AI breaches such as collusion, assessment outsourcing and cheating in exams were not evident.

学术诚信 (AI) 违规行为的增加导致高等教育机构实施解决方案以提高 AI 能力。有人认为，为了提高学生对人工智能的理解，应该在课堂层面教授概念和技能，并应考虑情境因素。本文对如何在一系列学科的课堂层面教授人工智能、情境因素如何影响人工智能教育方法以及这些方法如何与基于证据的建议保持一致进行了调查。采用有目的的抽样程序从澳大利亚两所大学的学科中选择研究单元。定性数据收集方法用于捕捉人工智能教育的方法，并通过扎根理论方法分析收集的数据。研究结果表明，人工智能主要通过明确的指导和个人讲故事来教授，并通过总结性评估进行评估。这种方法似乎受到个人哲学、机构要求和学生背景的影响。虽然这些方法符合最佳实践包括教育方法的概念，但为打击 AI 违规行为的增加而推广的最佳实践的其他方面并不明显，例如合谋、评估外包和考试作弊。