Discover, Embrace, Cultivate: A Reflective Approach to AI in Education
Steven Dawson, LaGuardia Community College
Bio
Steven Dawson is a NYC Department of Education educator and college advisor who supports MLLs through critical thinking, college readiness, and early college programming. He teaches courses in critical thinking and U.S. politics in partnership with LaGCC, where he focuses on helping students build independent thinking, academic confidence, and ethical AI literacy. His work centers on creating supportive pathways to higher education while helping students develop the skills needed to navigate complex information, systems, and ideas.
Course Setting
I teach in a NYC public high school and college partnership program serving primarily multilingual and immigrant students. My work takes place across both the high school and LaGuardia Community College, where students take college-level courses while receiving structured academic and advising support. I teach critical thinking and U.S. politics courses that focus on media literacy, independent thinking, and ethical AI use. Many students are first-generation college students navigating both language development and the transition into higher education simultaneously.
Problem of Practice
My problem of practice focuses on how to integrate AI into instruction in ways that support learning rather than replace thinking. As AI tools became more accessible, I noticed students increasingly relying on them to complete work without fully engaging with the reading, writing, or analytical process. This was especially important in my college-level courses serving multilingual learners, where authentic engagement and skill development are critical. In response, I began designing activities that require students to compare their own thinking with AI-generated responses, reflect on differences, revise intentionally, and develop greater awareness of AI’s strengths, limitations, and appropriate academic use. Focus Question: How might I design instruction and assignments that teach students to use AI as a support rather than a shortcut, while helping them produce work that reflects their actual engagement, understanding, and independent thinking?Strategy
I implemented a recurring instructional strategy built around three phases: students first complete a task independently, then compare their work to an AI-generated version, and finally reflect on the differences between the two. Tasks included summarizing readings, explaining political concepts, analyzing sources, and writing short responses. Students were required to submit both their original work and the AI response, then annotate or reflect on what the AI did well, where it lacked specificity, nuance, or evidence, and what revisions they would make to strengthen their own thinking.
To support accountability, students also documented the prompts they used and explained why they chose to use AI in a particular way. Over time, the goal was to move students away from passive copying and toward intentional evaluation and revision.
The rationale behind this strategy was grounded in metacognition and critical thinking research, particularly the idea that students learn more effectively when they analyze their own thinking processes and make conscious decisions about revision and evidence. Rather than banning AI, the strategy treated AI as an object of analysis, helping students develop judgment, skepticism, and awareness of the limitations of AI-generated responses.
Documentation
Making Thinking Visible Through AI Dialogue (JB)
This artifact captures a student engaging in a structured conversation with AI to clarify and deepen understanding of elections and political campaigns. It demonstrates how students used AI as a support tool for comprehension while practicing questioning, simplification, and evaluation of information rather than simply copying AI-generated responses.
Making Thinking Visible Through AI Dialogue (MR)
This artifact captures a student engaging in a structured conversation with AI to clarify and deepen understanding of elections and political campaigns. It demonstrates how students used AI as a support tool for comprehension while practicing questioning, simplification, and evaluation of information rather than simply copying AI-generated responses.
Measuring Impact
Overall Growth Analysis Across Assignments #1 (JB)
This data captures a student’s growth across multiple AI-supported assignments using two rubrics: metacognitive reflection quality and overall academic quality. Results demonstrate increased independent reasoning, stronger evidence integration, and more sophisticated evaluation of AI-generated responses over time, suggesting movement from passive AI use toward intentional and critical engagement.
Overall Growth Analysis Across Assignments #2 (MR)
This data captures a student’s progression across multiple AI-supported assignments using reflection and academic quality rubrics. Results demonstrate stronger analytical reasoning, improved synthesis of course materials, and increasingly critical evaluation of AI limitations. The student shifted from using AI for answer retrieval toward using it as a tool for clarification, inquiry, and deeper understanding.
Analysis
The strategy was successful in addressing my problem of practice by shifting students from passive AI use toward more reflective and intentional engagement. Across multiple assignments, students increasingly evaluated AI-generated responses rather than automatically accepting them as complete or authoritative.
To examine impact in depth, I selected two students as focal case studies because they were the most consistent in attendance, assignment completion, and reflection submission throughout the cycles. To supplement those case studies, I also reviewed four end-of-cycle student reflections and conducted qualitative coding to identify broader themes across the class. These reflections provided additional insight into student perceptions of AI use and helped contextualize the findings from the two focal students.
Student reflections showed growing awareness of AI’s strengths and limitations, particularly around overreliance, independent thinking, and critical evaluation. Qualitative coding of four student reflections revealed four recurring themes. All four students (4/4) reported attempting work independently before consulting AI. All four (4/4) described using AI as a support tool rather than a replacement for their thinking. All four (4/4) demonstrated increased awareness of when AI helps or hinders learning. All four (4/4) continued to wrestle with finding the boundary between productive support and overdependence. Representative comments included: “One thing I realized I can do without AI is start assignments by myself,” “I try not to copy directly because I learn more when I rework the ideas myself,” and “AI makes me think less when I depend on it too quickly without trying first on my own.”
Academic responses also improved over time, with stronger reasoning, better integration of course materials, and more independent analysis. Students increasingly described AI as a tool for organizing ideas, clarifying difficult concepts, and improving writing rather than as a source of answers.
One remaining challenge is ensuring students continue developing their own voice and reasoning rather than relying on AI for polished phrasing. Future versions will include more explicit revision requirements and opportunities for students to explain their reasoning verbally.
Recommendation
I would recommend this strategy to other CN and EC professors who are trying to navigate increasing student AI use without relying solely on restriction or detection. The strategy works especially well in reading- and writing-heavy courses where students must explain concepts, analyze sources, or develop arguments. Rather than treating AI as something separate from learning, the approach makes it part of the learning process itself.
A key factor is requiring students to compare their own thinking with AI-generated responses and reflect on the differences. This helps students develop judgment, skepticism, and metacognitive awareness while still benefiting from AI as a support tool. Repetition also matters. Using the same reflection structure across multiple assignments allowed students to become more analytical over time. For multilingual learners in particular, conversational AI interactions can support comprehension and confidence when paired with structured reflection and accountability.