Partnering with AI to create joy-centered personalized learning in a post-pandemic world

ABSTRACT

STEM instructors are teaching in classrooms shaped by wide variation in students’ prior preparation, learning experiences, and well-being, much of which has been intensified by the COVID-19 pandemic. At the same time, artificial intelligence (AI) tools are becoming increasingly available in higher education, prompting questions about how to use them productively and responsibly to support student learning. Personalized learning has been proposed as one approach for addressing these challenges, yet it remains difficult to implement at scale without increasing instructor workload or burnout. This Tips & Tools article presents a practical approach for partnering with AI to support joy-centered, personalized learning in STEM classrooms. Drawing on principles from joy-centered pedagogy, the procedure demonstrates how instructors can intentionally frame large language models (LLMs) to support student learning by fostering mattering, resilience, and agency through structured prompts. The approach formalizes instructional strategies implemented across lower- and upper-division undergraduate biology courses to support self-directed learning and growth-oriented revision, illustrating how AI can extend personalized learning opportunities at scale. The article introduces three prompt-based joy-centered pedagogy themes: normalizing uncertainty, promoting growth over perfection, and flattening classroom hierarchy. Each theme is accompanied by student-ready prompts and AI literacy guidance that instructors can readily integrate into courses. By positioning LLMs as pedagogical partners, this approach offers an accessible strategy for supporting diverse learners while preserving instructor bandwidth in an increasingly complex learning world.

INTRODUCTION

The rapid integration of artificial intelligence (AI) into higher education, within an already shaken post-pandemic learning landscape, calls for renewed attention to how teaching and learning are designed (1). While AI is disrupting education (2), the pedagogical challenges it brings into focus are not new. Instructors have long struggled with how to support diverse learners, sustain motivation, and foster meaningful engagement. These challenges have become more urgent in the wake of the COVID-19 pandemic, which further diversified students’ academic, social, and emotional needs (3, 4), widening gaps in academic preparation, particularly academic writing (5). Many students continue to experience lasting impacts on mental and emotional well-being (6), straining educators’ capacity to respond (7), and further intensifying an already inequitable academic system shaped by long-standing marginalization and socioeconomic barriers, particularly in STEM education (8).

To address the growing differences among students, personalized learning has been recognized as an approach that responds to students’ prior experiences and learning needs while supporting shared learning goals (9, 10). However, implementing personalized learning at scale has been challenging in higher education, characterized by large enrollments, standardized assessment structures, and growing instructor burnout (11, 12). Within this context, large language models (LLMs), an artificial intelligence system that uses deep learning to predict and generate human language, have gained traction as a tool for enabling personalized learning at scale (13–17).

In a moment defined by widespread student stress, mental health challenges, and social disconnection (18), personalization that focuses solely on cognitive differentiation risks overlooking the emotional and relational dimensions of learning. Joy-centered pedagogy, as articulated in Joy-Centered Pedagogy in Higher Education: Uplifting Teaching and Learning for All, offers a critical lens for addressing this gap; drawing from Camfield’s Chapter 1 definition, joy is understood as a renewable resource, an authentic, embodied, and connected state of uplift that instructors can intentionally cultivate to support students’ mattering, resilience, and agency (19). This article offers a practical, replicable procedure instructors can use to design joy-centered personalized learning experiences with LLMs in STEM classrooms.

PROCEDURE

Step 1: frame AI in joy-entered pedagogy

This procedure begins by framing joy-centered pedagogical values before using an LLM to support personalized learning. Using practice-oriented teaching strategies operationalized through the appendix to Joy-Centered Pedagogy: Uplifting Teaching and Learning for All (see Appendix: Teaching Tips to Spark Joy in Your Classroom [19]), instructors can provide a brief description of joy-centered pedagogy and select appendix themes or provide the full appendix and ask the system to foreground those values in student-facing responses. Because LLMs are highly responsive to prompt framing, this initial step helps ensure that LLM-generated interactions emphasize the desired engagement and support for the assignment. An example comparison illustrating how pedagogical framing shapes AI responses and supports personalized learning is provided in Appendix 1.

The rapidly evolving landscape of AI tools makes this approach intentionally flexible across platforms and institutional contexts. Instructors may have access to different tools depending on campus licensing agreements, student access, or personal preference. This procedure emphasizes prompt design and pedagogical framing as the primary influence of student experience. Instructors are encouraged to select tools that align with their context and to test prompts in advance, recognizing that outputs may vary slightly across systems. Importantly, instructors are encouraged to teach AI literacy skills. Appendix 2 provides instructor-facing guidance and sample student-facing language for discussing LLM limitations, hallucinations, bias, and student responsibility in biological reasoning.

Step 2: use structured prompts to promote personalized learning

Providing a common prompt set can create a shared starting point while allowing individualized engagement. The framing prompt provided in Appendix 1 and example follow-up prompts below are designed for instructors to share directly with students through course materials such as handouts, modules, or assignments. In this model, the instructor provides the pedagogical framing and structure, and students can copy/paste a prompt into an LLM and personalize it using their own contributions. The following joy-centered pedagogy prompt themes draw from the practice-oriented themes in the Appendix in Camfield 2025 (19) and were selected for their relevance in biology classrooms: normalizing uncertainty and productive struggle supports resilience (20); promoting growth over perfection supports mattering and resilience by countering fear of failure (21); and flattening hierarchy supports agency by positioning students as contributors to a shared learning community (22).

Theme 1: normalizing uncertainty and productive struggle

Why this matters

Biology learning requires integrating multiple representations and reasoning across scales, which can make uncertainty feel risky rather than productive. Many students disengage when confused rather than ask questions.

Instructor use

Use when students hesitate to ask questions or interpret confusion as failure.

Student-ready prompts

1. Name what’s unclear: I’m stuck on this topic: [topic]. Here is what I think I understand so far: [2–4 sentences]. Here is what feels unclear: [1–2 sentences]. Ask me three questions (one at a time) to help me locate what I’m missing. Please normalize confusion as part of learning.

2. Show me a starting point: I’m not sure how to begin this problem: [paste problem]. Suggest 2–3 reasonable ways to start and explain what each approach is trying to do. Don’t give the final answer yet.

3. Turn confusion into a good question: Help me turn my confusion into a question I can ask in class/office hours. My topic is: [topic]. What I’m confused about is: [describe]. Draft one version that sounds curious and confident.

These prompts create space for uncertainty, exploration, and confidence-building, key conditions for meaningful engagement in STEM learning.

Theme 2: promoting growth over perfection

Why this matters

Emphasis on correctness and high-stakes assessment can amplify perfectionism and fear of failure, limiting persistence and risk-taking.

Instructor use

Use when students fixate on grades, avoid revision, or disengage after mistakes.

Student-ready prompts

1. Focus on progress, not outcome: I’m working on this assignment or topic: [topic]. Help me identify what I’ve already learned or improved, even if my work isn’t complete.

2. Learn from a mistake: I made a mistake here: [paste or describe]. Help me understand what this mistake reveals about my thinking and how it can guide my next step.

3. Reframe revision: Here is my current draft or answer: [paste]. Identify one strength and one area for growth. Frame both as part of an ongoing learning process.

These prompts shift attention from performance to progress, helping students stay engaged and motivated as they build understanding in complex STEM contexts.

Theme 3: flattening hierarchy

Why this matters

Hierarchical STEM classrooms can discourage students from sharing ideas or seeing themselves as contributors to scientific sense-making.

Instructor use

Use when students defer to authority or hesitate to participate, especially during self-directed or open-ended learning activities.

Student-ready prompts

1. Design my learning: I have a self-directed learning assignment. Help me design a way to explore this topic that fits my interests or goals. Ask me questions to help me shape what I want to learn and why it matters to me.

2. Check my reasoning: Here is my current explanation: [paste]. Help me identify one part of my reasoning that seems well-supported and one part I should double-check using course materials or evidence.

3. Prepare to participate: Help me prepare one idea or question to contribute in class about [topic]. This is an invitation to think together.

These prompts foster a sense of agency and shared intellectual responsibility in STEM learning environments.

CONCLUSION

The procedure was implemented by the author across both lower- and upper-division undergraduate biology courses, ranging from 80 to 800 students at a large, public, minority-serving research institution. One application was an AI-supported “Pink Time” assignment, a self-directed learning activity in which one lecture was intentionally replaced with independent exploration aligned with course goals (23, 24) (Appendix 3). In a biodiversity course, students used this time to pursue personally meaningful biological inquiry, including species identification walks, biodiversity research, conservation-focused exploration, or creative projects. AI prompts helped students identify learning strategies, generate questions, and connect course content to personal contexts. The resulting range of student-directed applications reflected a breadth of personalized interest and creativity that would be difficult for a single instructor to support individually at scale. Instructor observations, along with recurring references to Pink Time in course evaluations and surveys, suggested that students experienced the assignment as a source of joy, agency, engagement, and ownership of learning.

This example offers an entry point for instructors exploring how AI might support joy-centered personalized learning in post-pandemic STEM classrooms. As educators adopt AI tools, transparent, opt-in practices and explicit conversations about responsible use remain essential (25) (Appendix 2). The strongest implementations will be those that adapt prompts to local contexts, invite student feedback, and use AI to support student agency, reflection, and joy in learning.

EDITOR'S NOTE

Ed. Note: This article was handled by Gregory J. Crowther, who acted as a guest editor in consultation with senior editor Samantha Parks.

SUPPLEMENTAL MATERIAL

The following material is available online at https://doi.org/10.1128/jmbe.00107-26.

Supplemental material. jmbe.00107-26-s0001.docx.

Appendices 1 to 3.

ASM does not own the copyrights to Supplemental Material that may be linked to, or accessed through, an article. The authors have granted ASM a non-exclusive, world-wide license to publish the Supplemental Material files. Please contact the corresponding author directly for reuse.