AI, and data-driven technologies in particular, have recently drawn attention to the importance of possessing an awareness and understanding of corresponding practices to engage with and participate in society i.e. to be data or AI literate. Education is an arena for promoting literacy and therefore teachers’ ability to provide teaching regarding these technologies is a prerequisite for equal participation. As policy and curricula are often nonspecific regarding digital and AI literacy, teachers require relevant scaffolding to interpret and enact curriculum changes. As such, this paper reflects on methods for eliciting teachers’ understanding and knowledge of AI to help effective scaffolding of teachers’ practices. Grounded in our ongoing empirical research we highlight challenges encountered in collecting and analysing data using different methods to understand teachers’ sense-making of AI. We find that certain methods, such as surveys, potentially fail to capture actual knowledge, understanding and attitudes towards ill-defined concepts such as AI. We conclude by discussing the potential implications of relying on data derived using certain methods and suggest alternative methods considering these limitations.