BlockTalk: GenAI Simulations for Conversational Debugging Practice
Paulina Haduong, Brian Yu, Miranda Shen, Bobby Gerami, Karen Brennan · Proceedings. · 2025
Conversing with students about challenges is central to the work of teaching computing through opportunities to create self-directed and open-ended programming projects.We introduce BlockTalk, a generative AI tool for teachers to practice conversational debugging through simulations.Our exploratory analysis examines teachers' interactions with the AI and offers insights for debugging pedagogy. Introduction and backgroundOpportunities to create self-directed and open-ended programming projects, where learners can bring their interests and identities into their computational creations, can broaden participation in computing (Tsan et al., 2023).Creating self-directed programming projects inevitably involves encountering and resolving bugs, and there is a need to understand how teachers support students in learning debugging practices.We designed a generative AI (GenAI) platform, BlockTalk, to support computing teachers' professional learning.On BlockTalk, teachers engage with conversational simulations to debug AI students' projects.We ask: How did computing teachers with varying experience interact with BlockTalk?The development of BlockTalk is informed by literature in three areas: (1) dialogic approaches to supporting student learning; (2) computing teachers' strategies for scaffolding student debugging; and (3) GenAI simulations for professional learning.Talking with students to scaffold their productive struggle can be part of a dialogic teaching repertoire (Alexander, 2018;Kim & Wilkinson, 2019).In computing education, teachers' approaches to scaffolding productive struggle have been found to vary along two dimensions: teacher versus student understanding and focus on the specific product versus student learning (Elliott et al., 2024).One way to support computing teachers' dialogic approaches can be through GenAI simulations, which can support teachers' need for more low-stakes and high-frequency practice spaces (Reich, 2022).