Source-linked AI summary
Adapting for AI: How elementary teachers adjust their practices for an AI-integrated curriculum
Fasika Melese, Ruiyang Wu, Xinyue Cui, Joanna Perkins, Xiaoyi Tian, Tiffany Barnes, Shiyan Jiang
TL;DR
Elementary teachers’ adaptation of student-facing AI curricula in real classrooms remains underexplored. This study follows teachers implementing a ToyTalk-based ELA and AI literacy curriculum and finds that their adaptive work spans technology, learner, and instructional tensions. The findings are bounded by a small, single-camp qualitative study centered on teachers’ accounts.
Problem
How elementary teachers adapt student-facing AI curricula in real classrooms remains underexplored, despite the importance of teacher work to implementation.
Source: §1 · p. 1Method
A qualitative case study followed three teachers implementing a ToyTalk-based ELA and AI literacy curriculum during a three-week summer camp.
Source: §3 · p. 3, §3.1 · p. 3Results
Teachers repaired technology, differentiated for learner readiness, translated abstract AI concepts through familiar stories, and balanced student agency with learning goals.
Source: §5.4 · p. 7Takeaways & Limitations
The findings frame teachers’ adaptive practices as repair, differentiation, translation, and balancing at the intersection of technology, learner, and instructional tensions.
Source: §5.4 · pp. 7–8The findings are snapshots from one three-week camp with few teachers and a reading-remediation student population, and rely on teachers’ accounts rather than objective classroom measures.
Source: §5.4 · p. 8
Abstract
from arXiv · showhide
Conversational AI tools are entering children's everyday experiences, and schools are interested in adopting them. However, successful classroom integration depends not only on the technology but also on the work teachers do to make it usable and appropriate for their students and classroom context. There is little known about how elementary teachers work as they implement conversational AI tools in real classrooms. In this study, we examine three teachers' experiences implementing an AI literacy and English Language Arts (ELA) curriculum built around ToyTalk, a conversational AI toy development platform, over 13 instructional days, a three-week summer camp. Drawing on daily individual reflections, group reflections, and post-camp interviews, we find that teachers' adaptive practices of repair, differentiation, translation, and balancing sit at the intersection of three tensions (technology, learner, and instruction). Teachers' understanding of AI and their role evolved over the camp experiences. From these findings, we contribute design implications and considerations for deploying conversational AI within elementary classrooms.
1 Introduction
Elementary teachers’ work is central to making student-facing conversational AI usable in real classrooms, where children’s developing literacy and computer skills shape implementation. This study examines how teachers adapt an ELA and AI literacy curriculum built around ToyTalk and the tensions they navigate.
1 Introduction
- Elementary learners’ developing reading, writing, and computer skills make classroom use dependent on teachers’ contextual support.Most AI literacy curricula were designed for secondary and higher education.Source: §1 · p. 1
- Teachers’ role in adapting student-facing conversational AI for elementary classrooms remains underexplored, especially as their role evolves during implementation.Prior work examined teacher-facing analytics and tutoring tools with older students.Source: §1 · p. 1
- The study follows three elementary teachers implementing a ToyTalk-based ELA and AI literacy curriculum after professional development.ToyTalk lets students customize conversational toys using pictures of physical toys and set their purpose, traits, tone, and rules.Source: §1 · p. 1
- The study asks how teachers adapt their role and what tensions they navigate while facilitating AI-mediated lessons.Source: §1 · pp. 1–2
- The paper contributes an empirical account of elementary teachers implementing AI curricula and design implications for student-facing tools that support teacher improvisation.Source: §1 · p. 2
2 Related work
Prior work frames teaching as situated adaptation, while research on AI in elementary education has largely emphasized student experiences, tool capabilities, or educators’ perceptions. This study addresses the limited evidence on teachers’ adaptation as they implement student-facing AI curricula over time.
2.1 Teachers’ situated practice, improvisation, and noticing
- Teaching involves improvising in response to classroom cues, and teachers collectively reshape mandated approaches through professional communities.Prior work calls this “disciplined improvisation” and describes AI teachers as “everyday repairers” [28, p. 2217].Source: §2.1 · p. 2
2.2 Chatbots and AI literacy for elementary students
- Research on elementary AI education reports students’ vague or misinformed conceptions and teachers’ limited computer science and computational thinking knowledge.Related work calls for co-designing curriculum with teachers, attending to students’ prior knowledge, and using relatable entry points.Source: §2.2 · p. 2
- Studies have examined children’s interactions with conversational agents and youth-created chatbots, but teacher mediation around student-facing AI remains underexplored.Existing work often centers student experience or tool capabilities, with few studies following teachers implementing specific curricula over time in elementary classrooms.Source: §2.2 · pp. 2–3
3 Method
The researchers conducted a qualitative case study of three teachers implementing an ELA and AI literacy curriculum with ToyTalk during a three-week summer camp. They analyzed teacher reflections, group debriefs, and interviews thematically.
3 Method
- A qualitative case study explored how elementary teachers made sense of and adapted their practice while implementing a ToyTalk-centered ELA and AI literacy curriculum.The exploratory approach was intended to provide contextual insight and inform hypotheses for later study.Source: §3 · p. 3
3.1 Study context
- The study took place in a three-week rural-district summer camp with 78 second- and third-grade students using a no-code platform to build conversational AI toys.The curriculum paired teacher-led ELA and AI learning sessions with ToyTalk.Source: §3.1 · p. 3
3.2 Participants
- Three non-CS teachers with limited prior experience integrating AI taught students from six elementary schools in a rural district.Source: §3.2 · p. 3
3.3 Data Collection
- Data included daily individual reflections, daily group debrief sessions, and post-camp semistructured interviews with each teacher.Teachers completed an average of 8–12 reflection entries; group debriefs lasted 90 minutes, and interviews lasted 45–60 minutes.Source: §3.3 · p. 3
3.4 Data Analysis
- The first author inductively coded the reflections and interviews, then collaboratively refined emerging patterns into seven themes.The analysis identified 13 codes across the three data sources.Source: §3.4 · p. 4
4.1 RQ1: How do teachers adapt their role as they implement ELA+AI literacy curriculum?
Teachers adapted their roles through flexibility, improvisation, and strategies responsive to students’ needs. Across the camp, they addressed technical and readiness challenges, adjusted instruction, and developed their understanding of AI and their teaching role.
4.1 RQ1: How do teachers adapt their role as they implement ELA+AI literacy curriculum?
- Teachers adapted their roles by modifying lessons and developing strategies to meet students’ needs.Source: §4.1 · p. 4
4.1.1 Initial challenges.
- Early challenges included microphone errors, system outages, student engagement, and pacing difficulties, particularly for lower-performing students.Teachers also found that students’ broad prompts could produce generic or incorrect answers; one teacher modeled making a question about a story more specific.Source: §4.1.1 · p. 4
- Prefilled ToyTalk templates improved response consistency with the intended topic, including when students asked off-topic questions.Students came to understand that people can shape AI behavior through rules and guidelines.Source: §4.1.1 · p. 4
- Teachers often split lessons across days, especially for second graders, to allow deeper exploration and accommodate different learning paces.Teachers reported that some lessons were rushed or unfinished within a single day.Source: §4.1.1 · p. 4
4.1.2 Pedagogical adaptations and strategies.
- Teachers differentiated support for students who struggled with writing, using teacher transcription, drawing or emojis, and peer assistance from older or more proficient students.Source: §4.1.2 · p. 5
- Hands-on activities, including pattern-recognition bracelets and movement breaks, made abstract concepts more tangible and supported engagement.Source: §4.1.2 · p. 5
- Teachers modeled prompt-writing through an “I do, we do, you do” sequence to gradually release responsibility to students.Source: §4.1.2 · p. 5
- Teachers connected AI to familiar narratives and broader themes, while recognizing that students’ reading and computer skills affected their ability to engage with the curriculum.They linked AI errors to “The Boy Who Cried Wolf” and discussed trust, critical thinking, and technology careers.Source: §4.1.2 · p. 5, §4.1.3 · p. 5
4.2 RQ2: What tensions do teachers navigate when facilitating AI-powered tool mediated lesson plans?
Teachers navigated tensions involving technology reliability, diverse learner needs, curriculum adaptability, and their own preparedness. These tensions required improvisation and pedagogical adjustment.
4.2 RQ2: What tensions do teachers navigate when facilitating AI-powered tool mediated lesson plans?
- Teachers navigated tensions involving technology reliability, diverse learner needs, curriculum adaptability, and their own preparedness.These tensions required improvisation and pedagogical adjustments.Source: §4.2 · p. 5, §4.2.4 · p. 6
4.2.1 Technical issues and their impact.
- Microphone failures, slow responses, and error messages disrupted tool use, frustrating students with speech impediments and requiring attention to speech-to-text errors.A speech-to-text system sometimes changed “pear” to “pair,” and students did not always reread the input.Source: §4.2.1 · p. 6
4.2.2 Addressing diverse students needs.
- Students differed in computer literacy, language, reading comprehension, and engagement, while basic login, typing, handwriting, and instruction-reading difficulties consumed instructional time.Teachers also weighed writing practice against the risk of student frustration and disengagement; worksheets and exit tickets challenged students with limited writing skills.Source: §4.2.2 · p. 6
4.2.3 Curriculum design and adaptability.
- Lessons often ran too long for second graders, while rapid introductions to complex AI concepts, independent worksheets, and open-ended toy design required teachers to adjust curriculum structure.Teachers proposed more scaffolding, including modeling, pre-filled templates, and clearer alignment between creative choices and learning objectives.Source: §4.2.3 · p. 6
4.2.4 Teacher preparedness and support.
- Teachers were new to AI concepts and stressed by materials finalized shortly before implementation, but valued collaborative briefings where they could share challenges and suggest changes.The iterative feedback loop helped address emerging technological, learner, and instructional tensions.Source: §4.2.4 · p. 6
5 Discussion
The discussion frames teachers’ adaptive work as a response to tensions among technology, learners, and instruction. It presents four practices, collective sense making, and design implications, while noting the study’s limited scope.
5.1 Teachers adaptive practice in AI-integrated curriculum
- Teachers’ repair, differentiation, translation, and balancing practices address tensions among technology, learners, and instruction, extending prior orchestration accounts to generative AI tools.Repair supplies robustness when technology fails; differentiation responds to student cues; translation makes AI concepts accessible; balancing negotiates learning goals with student agency.Source: §5.1 · p. 7, §5.4 · p. 8
5.2 Sense making as a collective
- Teachers described learning AI alongside students, while group reflections let them raise problems, propose modifications, and feel heard as they built technological understanding in real time.Their emerging AI-specific knowledge built on pedagogical and content expertise.Source: §5.2 · p. 7
5.3 Design implications
- Because adaptation is described as implementation’s mechanism, AI curricula and student-facing tools should support teacher repair, foundational-skill scaffolds, conceptual translation, and structured reflection.Suggested supports include manual fallbacks, multimodal input and writing templates, ready analogies and examples, and daily individual and group reflection channels.Source: §5.3 · p. 7
5.4 Conclusion
- Across a three-week camp, teachers repaired technology, differentiated for learner readiness, translated abstract AI concepts through familiar stories, and balanced creative agency with learning goals.Teachers responsively redesigned the curriculum to fit students’ needs; the resulting framework identifies four adaptive practices at the intersection of three tensions.Source: §5.4 · pp. 7–8
- The findings are snapshots from one three-week prototyping camp with few teachers and students in a reading-remediation program, centered on teachers’ accounts rather than objective classroom measures.The authors call for study across multiple sites and a full school year, including student perspectives and tests of whether the four practices apply to other AI curricula and tools.Source: §5.4 · p. 8
A.1 Interview Protocol
The interview protocol asked teachers to reflect on their experience with AI, decisions made during camp, and how they might adapt the curriculum for future implementation.
A.1 Interview Protocol
- Teachers were asked about prior classroom experience with AI tools and a specific moment when they made an unplanned decision during camp.Follow-up prompts asked what they noticed and what knowledge or experience informed the decision.Source: §A.1 · p. 10
- The protocol asked teachers to consider what physical or embodied experiences offer beyond screen-based activities and what hands-on components they would add..Source: §A.1 · p. 10
- Teachers were asked about tensions between following lesson plans and meeting student needs, plus what should change for school-year use and teacher preparation.Prompts covered realistic timing, duration, curriculum integration, and professional development support.Source: §A.1 · p. 10
A.2 Daily reflection survey prompts
Daily reflection prompts asked teachers to identify the day’s AI learning, what went well or could improve, and notable student-created AI toys.
A.2 Daily reflection survey prompts
- Teachers were asked to name the most important AI idea explored that day..Source: §A.2 · p. 10
- Prompts asked teachers to record what went well and what could be improved in the day’s lesson..Source: §A.2 · p. 10
- Teachers were invited to report inspiring moments and describe AI toys students made, including creations that surprised or stood out to the class..Source: §A.2 · p. 10