Time for SITE 2025 in Orlando! You can find all the resources, slides, etc. from my sessions here
Monday, March 17
Workshop: AI as a (Uniquely Useful) Cognitive Illusion
Tuesday, March 18
Analyzing AI-Generated Feedback
Melissa Warr and Nicole Oster
This paper examines potential biases in AI-generated feedback on student writing, focusing on how ChatGPT 4.0 adjusts feedback for students identified as Hispanic. Utilizing Linguistic Inquiry and Word Count (LIWC) and Critical Discourse Analysis (CDA), this study investigates differences in feedback discourse, examining the potential impact on diverse student populations and the risk of exacerbating existing educational inequalities. Findings reveal ChatGPT provides feedback that reflects higher levels of clout to Hispanic students, suggesting variations in authority and expertise in the language used. These patterns may reflect and influence broader socio-historical conditions, raising concerns about the reinforcement of existing educational disparities. This study underscores the need for critical evaluation of AI in educational settings to prevent the magnification of biases and ensure equitable educational experiences.
Embedding and Developing TPACK with LLM Prompt Writing
Melissa Warr and Katherinne Bardales Sardena
In this paper, we explore the integration and development of Technological Pedagogical Content Knowledge (TPACK) through the creation of prompts for large language models (LLMs), through the development of a translanguaging chatbot. The study investigates how designing LLM prompts can both benefit from and enhance the designer’s TPACK. By analyzing the process of developing a chatbot prompt to support bilingual students in elementary schools, we identify elements of TPACK in the interactions between the prompt designer and the LLM. The findings suggest that constructing LLM prompts may serve as a valuable tool for pre- and in-service teachers, not only in their teaching practices but also in developing the TPACK necessary for effective and critical use of AI in education.
Wednesday, March 19
AI-Integrated Design for Educators
Lindsey McCaleb, Punya Mishra, Nicole Oster, Melissa Warr
The emergence of Generative AI marks another significant technological advancement that has the potential to revolutionize education. However its integration into teaching and learning requires a nuanced understanding of its possibilities, limitations, and implications. In this paper, we discuss the unique properties of AI and key principles intended to foster inquiry around using AI in education. We introduce the AI-Integrated Design for Education (AIDE) framework, describing key principles in detail, exploring ethical considerations, and provide examples of AIDE in practice through collaborative design processes. AIDE builds on the idea of GenAI as being a “smart drunk (occasionally biased) intern” – incredibly capable and creative, yet sometimes erratic and prone to mistakes and biases. As Koehler et al. (2011) discuss using TPACK with teachers, they encourage the use of “deep play,” an engagement with rich problems of pedagogy, technology, and content, and their inter-relationships. By incorporating “deep play” in professional development, educators can be inspired to use the process of design with their students (Koehler et al, 2011). The goal is to empower educators to approach AI as a collaborative tool that enhances their expertise rather than replaces it, leveraging AI’s strengths while compensating for its weaknesses through human judgment and expertise.
- Paper
- Slides
Applications of AI in K-12 and Higher Education: Quickfires for AI Use in Every Subject
Melissa Warr, Punya Mishra
We know very little about what generative AI tools can do, and it is only through our own experimentations that we can understand and fully utilize these technologies. Furthermore, experimentation brings greater familiarity and facility with AI tools. In this presentation, we will share “quickfires” for AI: creative AI challenges for learners to attempt in a limited time period. Learners are likely to fail the task (and they are aware of this dynamic), lowering expectations and enabling more creative and risky experimentation. For example, a quickfire may ask learners to produce an interactive simulation of the periodic table, create a complete budget for a new business, or represent geometric figures through music. Through interactions with AI tools, learners come to better understand these tools in disciplinary contexts while developing dispositions for exploration and innovation.
Thursday, March 20
Challenging AI Bias through Critical Pedagogy
Melissa Warr
In this paper, I explore progress in research on the intersection of Generative Artificial Intelligence (GenAI) and critical pedagogy, advocating for an equity-oriented approach to AI in education. I describe four guidelines for critical AI use: active engagement, continual reflection, collaborative exploration, and creative discovery. Then I describe two contexts in which I am studying how these guidelines impact AI use: a study with pre-service teachers, and a second study with local in-service teachers that includes teachers designing curriculum for other teachers.
Approach Leveraging LLMs in Supporting PCK Development
Suparna Chatterjee, Melissa Warr
The role-playing abilities of Large Language Models (LLMs) prompted us to investigate their potential in developing pedagogical content knowledge (PCK) for teacher candidates. We engaged an LLM as a third-grade student in mathematics and science scenarios, guiding it toward new conceptual understandings. These role-play interactions deepened our exploration of the relationship between content and learner cognition. We initiated the first stage of an Educational Design Research (EDR) study to address two key questions for (1) Intervention exploration: What design principles facilitate PCK development using LLMs? and (2) Theoretical Understanding: What types of thinking and connections emerge in this process? Findings highlight three themes: LLMs enhance thought processes, feedback, and content explanation; reflection before, during, and after interactions fosters improvement; and scaffolding is essential for productive engagement. These insights inform three design principles—structure, modeling, and reflection—guiding future research investigating the second stage of EDR for co-exploration with teacher educators to refine theoretical understanding and apply design principles in instructional design.
- Paper
- Slides