Mutual trust between teachers and students is a prerequisite for effective teaching, learning, and assessment in higher education. Accurate predictions about the other group's use of generative artificial intelligence (AI) are fundamental for such trust. However, the disruptive rise of AI has transformed academic work practices, raising important questions about how teachers and students use these tools and how well they can estimate each other's usage. While the frequency of use is well studied, little is known about how AI is used, and comparisons with similar practices are rare. This study surveyed German university teachers (N = 113) and students (N = 123) on the frequency of AI use and the degree of delegation across six identical academic tasks. Participants also provided incentivized cross-sample predictions of the other group's AI use to assess the accuracy of their predictions. We find that students reported higher use of AI and greater delegation than teachers. Both groups significantly overestimated the other group's use, with teachers predicting very frequent use and high delegation by students, and students assuming teachers use AI similarly to themselves. These findings reveal a perception gap between teachers' and students' expectations and actual AI use. Such gaps may hinder trust and effective collaboration, underscoring the need for open dialogue about AI practices in academia and for policies that support the equitable and transparent integration of AI tools in higher education.

翻译：师生间的相互信任是高等教育中有效教学、学习与评估的前提。对另一方群体使用生成式人工智能（AI）的准确预测是这种信任的基础。然而，AI的颠覆性崛起已改变了学术工作实践，引发了关于师生如何使用这些工具以及他们能在多大程度上准确估计对方使用情况的重要问题。尽管使用频率已得到充分研究，但关于AI如何被使用却知之甚少，且与类似实践的对比研究更为罕见。本研究调查了德国大学教师（N = 113）和学生（N = 123），针对六项相同的学术任务，收集了其AI使用频率及委托程度的数据。参与者还提供了有激励的跨样本预测，以评估其对另一群体AI使用情况预测的准确性。我们发现，学生报告的AI使用频率和委托程度均高于教师。两组群体均显著高估了对方的使用情况：教师预测学生使用非常频繁且委托程度高，而学生则假定教师与自己的使用方式相似。这些发现揭示了师生对AI使用的期望与实际行为之间存在认知差距。此类差距可能阻碍信任与有效协作，凸显了在学术界开展关于AI实践的公开对话、以及制定支持AI工具在高等教育中公平透明整合的政策的必要性。