Teacher noticing has become increasingly acknowledged as a fundamental aspect of teacher professional competence. Teacher education scholars have examined how the development of noticing might be supported both in initial teacher education and in professional development. In mathematics teacher education, several studies have explored the use of video as a supporting tool for teacher noticing. It remains unclear how this body of work builds on the various theoretical perspectives of noticing prevalent in the literature, thus broadening our understanding of noticing. Furthermore, the field has not examined systematically the extent to which research has leveraged the affordances of digital video technologies, and whether scholars have employed different research methods to answer questions that are critical to teacher educators. This survey paper reviews studies published in the last two decades on programs centered on mathematics teacher noticing that used video as a supporting tool for teacher learning. Thirty-five peer-reviewed papers written in English were identified and coded along three dimensions: (1) theoretical perspectives; (2) use of video technologies; and (3) research questions and methods. This review summarizes important findings and highlights several directions for future research. Most studies involved pre-service teachers, and only a few centered on in-service teachers. Developers of the large majority of programs took a cognitive psychological perspective and focused on the attending/perceiving and interpreting/reasoning facets of noticing. Few studies used video-based software and few studies used grouping, and even fewer used randomized grouping. Evidence of program effects on responding and decision making, and on instructional practice, is limited and should be extended in the future.

We present reflections on our experience using a generative AI model in qualitative research, to illuminate the AI's contributions to our analytic process. Our analytic focus was a segment of classroom transcript, which captured a teacher introducing scientific theory-building practices to middle school students. We used a grounded theory approach to produce a fine-grained characterization of the teacher's talk moves during the lesson implementation. Our eventual goal is to build a more nuanced conceptualization of responsive teaching in the context of theory-building activities. We involved GPT-4 during the initial exploratory and later focused coding stages. For our analysis of GPT-4’s contributions to the analytic process, we analyzed our notes and analytic memos, along with video recordings of meetings where we discussed insights in response to GPT-4’s input. We present vignettes to illustrate pivotal moments where AI contributed to the coding process, including code generation, comparison, and refinement. The paper presents our experiences of conducting qualitative research in partnership with generative AI, underscoring the role that emerging technologies can play in the analysis of data and the development of grounded theory.

Large language models (LLMs) have shown promise in generating sensible code explanation and feedback in programming exercises. In this experience report, we discuss the process of using one of these models (OpenAI's GPT-4) to generate individualized feedback for students' Java code and pseudocode. We instructed GPT-4 to generate feedback for 113 submissions to four programming problems in an Algorithms and Data Structures class. We prompted the model with example feedback (few-shot learning) and instruction to (1) give feedback on conceptual understanding, syntax, and time complexity, and (2) suggest follow-up actions based on students' code or provide guiding questions. Overall, GPT-4 provided accurate feedback and successfully built on students' ideas in most submissions. Human evaluators (computer science instructors and tutors) rated GPT-4's hints as useful in guiding students' next steps. Model performance varied with programming problems but not submission quality. We reflect on where the model performed well and fell short, and discuss the potential of integrating LLM-generated, individualized feedback into computer science instruction.