Exploring Teacher-Chatbot Interaction and Affect in Block-Based Programming DOI: XXXXXXX.XXXXXXX Conference: The ACM CHI conference on Human Factors in Computing Systems; April 13–17, 2018; Barcelona, Spain ISBN: 978-1-4503-XXXX-X/2026/04 Conference: Proceedings of the 2026 CHI Conference on Human Factors in Computing Systems; April 13–17, 2026; Barcelona, Spain Proceedings of the 2026 CHI Conference on Human Factors in Computing Systems (CHI ’26), April 13–17, 2026, Barcelona, Spain DOI: 10.1145/3772318.3791823 ISBN: 979-8-4007-2278-3/2026/04 CCS: K-12 Education CCS: Human Computer Interaction CCS: Natural Language Generation
Bahare Riahi, Ally Limke, Xiaoyi Tian, Viktoriia Storozhevykh, Sayali Patukale, Tahreem Yasir, Khushbu Singh, Jennifer Chiu, Nicholas lytle, Tiffany Barnes, Veronica Cateté · 2026
Read Exploring Teacher-Chatbot Interaction and Affect in Block-Based Programming DOI: XXXXXXX.XXXXXXX Conference: The ACM CHI conference on Human Factors in Computing Systems; April 13–17, 2018; Barcelona, Spain ISBN: 978-1-4503-XXXX-X/2026/04 Conference: Proceedings of the 2026 CHI Conference on Human Factors in Computing Systems; April 13–17, 2026; Barcelona, Spain Proceedings of the 2026 CHI Conference on Human Factors in Computing Systems (CHI ’26), April 13–17, 2026, Barcelona, Spain DOI: 10.1145/3772318.3791823 ISBN: 979-8-4007-2278-3/2026/04 CCS: K-12 Education CCS: Human Computer Interaction CCS: Natural Language Generation by Bahare Riahi, Ally Limke, Xiaoyi Tian, Viktoriia Storozhevykh, Sayali Patukale, Tahreem Yasir, Khushbu Singh, Jennifer Chiu, Nicholas lytle, Tiffany Barnes, Veronica Cateté (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-20211.txt.
Introduction
Riahi and co-authors study how middle school teachers interact with chatbots during a coding activity. Eleven teacher teams used chatbots inside a block-based programming setting for science. Audio transcripts and chat logs feed a qualitative analysis. Three teacher profiles emerge: explorer, frustrated, and mixed, each needing different support. Teachers praised gains in prompting skill and confidence. They also feared losses in learning and critical thinking for students. Design advice covers staged chatbot introductions and teacher control of features. Guidance on when to use chatbots rounds out the findings.
Catalog record
| CREATOR | Bahare Riahi, Ally Limke, Xiaoyi Tian, Viktoriia Storozhevykh, Sayali Patukale, Tahreem Yasir, Khushbu Singh, Jennifer Chiu, Nicholas lytle, Tiffany Barnes, Veronica Cateté |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 90208 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2603.20211
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Exploring Teacher-Chatbot Interaction and Affect in Block-Based Programming DOI: XXXXXXX.XXXXXXX Conference: The ACM CHI conference on Human Factors in Computing Systems; April 13–17, 2018; Barcelona, Spain ISBN: 978-1-4503-XXXX-X/2026/04 Conference: Proceedings of the 2026 CHI Conference on Human Factors in Computing Systems; April 13–17, 2026; Barcelona, Spain Proceedings of the 2026 CHI Conference on Human Factors in Computing Systems (CHI ’26), April 13–17, 2026, Barcelona, Spain DOI: 10.1145/3772318.3791823 ISBN: 979-8-4007-2278-3/2026/04 CCS: K-12 Education CCS: Human Computer Interaction CCS: Natural Language Generation Bahare Riahi, Ally Limke, Xiaoyi Tian, Viktoriia Storozhevykh, Sayali Patukale, Tahreem Yasir, Khushbu Singh, Jennifer Chiu, Nicholas lytle, Tiffany Barnes, Veronica Cateté Abstract . AI-based chatbots have the potential to accelerate learning and teaching, but may also have counterproductive consequences without thoughtful design and scaffolding. To better understand teachers’ perspectives on large language model (LLM) based chatbots, we conducted a study with 11 teams of middle-school teachers using chatbots for a science and computational thinking activity within a block-based programming environment. Based on a qualitative analysis of audio transcripts and chatbot interactions, we propose three profiles: explorer, frustrated, and mixed that reflect diverse scaffolding needs. In their discussions, we found that teachers perceived chatbot benefits such as building prompting skills and self confidence alongside risks including potential declines in learning and critical thinking. Key design recommendations include scaffolding the introduction to chatbots, facilitating teacher control of chatbot features, and suggesting when and how chatbots should be used. Our contribution informs the design of chatbots to support teachers and learners in middle school coding activities. † † cc-license: by 1. Introduction While integrating Artificial Intelligence (AI) applications in K-12 education has had a long and evolving history, the introduction of ChatGPT and other Large Language Model (LLM) powered technologies in 2022 accelerated the adoption of these tools ( 26 ; 43 ) . This rapid proliferation has created a litany of pedagogical and ethical concerns about problematic AI-usage in the classroom, with many stemming from uncertainty about how much of graded material (e.g., homework, assignments, exam answers) was completed by these LLMs rather than by the students themselves ( 29 ) . Products designed to detect LLM-generated text are not without error ( 79 ) , and there are inconsistencies at the school, course, and teacher-level about what types of LLM interactions constitute plagiarism ( 58 ) . For instance, a student generating small sections of code or test cases for a larger coding project may or may not be seen as acceptable. Additionally, most of these AI-interactions are student-self directed and in environments outside of either the visibility or control of the teacher. This type of use and misuse of AI in classrooms can potentially have impacts on learning outcomes. While our understanding of these effects are still evolving, early evidence points to the idea that while unrestricted student use can lead to faster and more accurate completion of activities in the short term, there can be long-term learning consequences. Research points to the idea that offloading cognitive effort to an AI may reduce the mental struggle necessary for deep learning, particularly in domains like programming where "productive struggle" is a key component of skill acquisition ( 14 ) . When students rely on an LLM to generate code, they may bypass the critical thinking and problem-solving processes that lead to lasting understanding ( 27 ) . Companies are responding to these challenges by creating LLM systems targeted towards more effective pedagogical practices (e.g. ChatGPTs StudyMode ( 60 ) ). Professional development workshops will allow researchers and chatbot designers to capture the range of concerns teachers have about unwanted interactions with the chatbot as well as set standards for desired AI behaviors for both student-AI and teacher-AI interactions. Additionally, the range of learning environments and activities where these chatbots can be employed and differences in teachers’ pedagogical style suggest that a ‘one-size-fits-all’ system with no instructor control over behavior is unacceptable. These complexities highlight the need for psychological frameworks such as Technology Acceptance Models (TAM) ( 19 ) and Theory of Planned Behavior (TPB) ( 2 ) to explain teachers’ evaluations and adoption decisions, and we mapped our analysis to these frameworks to interpret how their emotions, perceptions, and sense of control shaped their intention to use the chatbot. In the summer of 2025, we invited 25 middle school science teachers to participate in a professional development focused on integrating computational thinking (CT) and block-based programming activities into their science classrooms. As part of this training, we included a module in which teachers completed a lesson within a block-based environment that had an integrated ‘chatbot’ component. This chatbot had a range of features including the generation of code from written description as well as the ability to answer questions regarding programming concepts generally or questions related to how to use the block-based environment specifically. Our aim was to use data collected from both these learning modules and a set of follow up discussions with the teachers to guide our understanding of how teachers interact with these types of technologies in practice, and how teachers would want a more formally developed AI system to operate. This aim is characterized by the following research questions: Research Questions: (1) RQ1. How does interacting with chatbots for learning programming impact middle school teachers’ affect and attitudes? (2) RQ2. What are teachers’ perspectives about the benefits and risks of using an LLM for block-based programming in their classrooms for students and for teachers? 2. Related Work 2.1. How Teachers Interact with LLMs in Pedagogical Environments Kim et al. (2024) conceptualized teacher AI interactions as cognitive, socio-emotional, and artifact-mediated ( 46 ) . Cognitive interactions involve teachers’ reasoning about AI outputs, selecting or modifying tasks, and making instructional decisions. Socio-emotional interactions reflect the trust, confidence, and attitudes of teachers toward AI, influencing when and how they use AI suggestions. Artifact-mediated interactions occur when teachers engage iteratively with AI-generated outputs, treating them as artifacts to support lesson design, instructional decision-making, and student learning. This framework, applied to various AI tools and classroom dashboards, can also be observed in studies examining the use of LLMs, such as ChatGPT. Holstein et al. (2018) evaluated Lumilo, a wearable mixed reality tool for K-12 teachers, providing real-time insights into student learning and behavior ( 33 ) . Teachers engaged cognitively by planning interventions, socio-emotionally by evaluating trust and usability, and artifact-mediated through learning analytics. Lumilo also highlighted challenges with LLM use, such as information overload, anxiety, and the need to balance interpretability with accuracy. Across these studies, teachers’ interactions with LLMs and AI tools reflect cognitive, socio-emotional, and artifact-mediated dimensions. Teachers use AI outputs to reason, adapt, or reject; navigate trust in AI; and iteratively interact with AI-generated artifacts to shape instruction. They prioritize maintaining agency, seeking tools that reduce burdens while preserving pedagogical control and ethical responsibility ( 82 ; 24 ) . Overall, effective teacher–LLM interactions are best seen as collaborative partnerships where AI augments, not replaces, teacher expertise. LLMs and generative AI have quickly entered schools, generating excitement and concerns. Teachers view ChatGPT as a valuable tool for saving time and supporting computer science lesson planning ( 62 ) , but also recognize challenges like academic integrity, student