In the public education system of China, intelligent multimedia classrooms equipped with camera-based systems are being progressively introduced to capture and quantify teacher-student interactions. These systems generate digital reports with behavioral ratios intended to help teachers reflect on instructional design and interaction patterns, which align with the functions of Flanders Interaction Analysis (FIA), a classic classroom observation tool (Amatari, 2015). However, teachers may feel pressured to “teach to the metrics,” prioritizing data-friendly but musically superficial interactions over pedagogically meaningful engagement (Heiter, 2016). Meanwhile, Duke (1999) critically states that existing research disproportionately focuses on observable teacher behaviors, implicitly equating them with effective student learning outcomes, and notes that this design inherently simplifies the complex, multidimensional flow of music classroom interactions. As FIA privileges verbal behaviour as the most reliable and representative indicator of classroom dynamics and explicitly excludes non-verbal gestures (Amatari, 2015), such technology-driven data measurement may be misleading in a discipline characterized by diverse forms of participation and deeply culturally situated practices, such as music.The purpose of this study is to investigate how data-driven assessment of teaching behaviors, enabled by intelligent technology, may influence the curriculum design and professional identity of general music teachers in Chinese public schools. This qualitative case study (Stake, 1995), informed by a critical theory lens (Matsunobu & Bresler, 2014), conducted semi-structured interviews (Rubin & Rubin, 2012) with five music teachers working in a public school equipped with multimedia classrooms and a data-driven classroom observation system, and performed a document analysis of their curriculum-related data reports. Examining teacher responses to automated data. Open and focused coding (Saldaña, 2013) was used to analyze the data.Preliminary findings suggest that data-driven teaching behavior assessment systems may potentially shape in-service music teachers’ perceptions of “quality music teaching” and further influence their practical orientation as music educators. This may lead to a shift in teachers’ curriculum design focus—from being “student-centred” towards pursuing improved “data presentation.” As a result, key musical literacy activities, such as improvisation, are marginalized because they are difficult to capture through verbal and behavioral data. Secondly, for music teachers’ identity, when pedagogically sound practices are not recognized by the simplified technological metrics, teachers may face the risk of ultimately undermining their teaching autonomy and the development of a creative professional identity.In this presentation, I explicate these findings to draw the attention of educational administrators and policymakers to the potential limitations of “technology-driven” teaching assessment models and their associated risks—particularly in arts education, a discipline imbued with complex socio-cultural values and creative interactions.