.png)
AI Is Not the Enemy—It’s a Thought Partner That Can Revolutionize Teaching and Learning
- Genevieve Jomantas
- Feb 2
- 9 min read
By Genevieve Jomantas M.Ed, M.S.
Introduction
Since AI tools like ChatGPT exploded into higher education in November 2022, conversations about academic integrity, plagiarism detection, and AI misuse have taken center stage. Companies like Turnitin have rushed to create AI detection tools to catch students who use AI to generate their work. But here’s the problem—these AI detectors are unreliable, inaccurate, and discriminatory.
According to Stanford’s Institute for Human-Centered AI (2023), AI detectors are far more likely to flag writing from non-native English speakers as AI-generated—even when the work is entirely original. Let that sink in. Students who already face linguistic challenges are now at an even higher risk of being falsely accused of academic dishonesty simply because of how they write. That’s not just a technological flaw—it’s an ethical disaster.
One of the biggest reasons AI detection fails is that ChatGPT is designed to produce human-like responses. Some people claim they can recognize ChatGPT-generated text by spotting overused words like "delve" or noticing a lack of emotional nuance and repetitive phrasing.
However, with tools like Grammarly and other AI-assisted revision programs, students can refine AI-generated text to sound more natural, making detection nearly impossible.
With AI rapidly changing how we problem-solve and create, it’s time for educators to stop fighting AI and start exploring how it can enhance teaching and learning. This is precisely what Chris Mah and Sarah Levine discuss in their podcast and blog post, How to Use ChatGPT as an Example Machine (Cult of Pedagogy, 2023). This post expands on their ideas and explores how AI can support student learning through Cognitive Load Theory (CLT).
AI as an Example Machine: How ChatGPT Enhances Learning
According to Mah and Levine (2023), students learn best when exposed to multiple concept examples. The challenge? Teachers often work with students at vastly different skill levels and learning experiences. Generating enough examples that cater to everyone’s needs and facilitate efficient and effective learning is arduous, if not impossible. The higher education landscape has drastically changed over the last twenty years in terms of not only educational technology but the learners we serve.
In a blog post, the higher education data analytics company Watermark Insights highlighted that today's student body is no longer homogenous and that how students want to learn is evolving (Watermark Insights, n.d.). Traditional higher education models, which assumed a standard student experience, are becoming outdated as institutions serve an increasingly diverse population that balances work, family, and education. This shift in student demographics and needs requires a fundamental change in how colleges approach teaching and learning.
Bruce Mackh, in Pivoting Your Instruction (2021), emphasizes that colleges should transition from expecting students to be "college-ready" to becoming "student-ready." Historically, higher education has been faculty-centered, admitting only those who are academically prepared. Instead of requiring students to conform to outdated systems, institutions should create structured, inclusive learning experiences that cater to diverse student needs. The question remains: how can institutions achieve this?
One practical approach is to integrate AI literacy into first-year programs. A 30-minute exercise on prompt engineering, as discussed by Mah and Levine, can empower students to effectively use AI tools like ChatGPT. By learning to craft precise prompts, students can receive personalized support tailored to their individual learning needs. This not only enhances their educational experience but also aligns with the goal of creating a student-ready institution.
Mah and Levine demonstrate how ChatGPT can generate dynamic, adaptable examples, allowing students to engage with the material more deeply. The key takeaway? Never settle for ChatGPT’s first response. Instead, students and educators should refine, iterate, and tailor AI-generated outputs to enhance understanding and critical thinking. This approach lies at the heart of a practical learning experience that prioritizes inclusivity, alignment with learning goals, and meaningful engagement. The more we actively interact with AI, the better we can harness its potential as a powerful educational tool rather than just a content generator.
Cognitive Load Theory: Why AI is a Game-Changer for Teaching and Learning
When designing instruction, educators must consider Cognitive Load Theory (CLT)—a framework that explains how the brain processes and retains information (Sweller, 1988). Understanding cognitive load isn’t just theoretical; it directly impacts how we teach and students learn. With AI tools like ChatGPT, we now have new ways to enhance learning, reduce unnecessary cognitive strain, and make education more accessible to all students.
The Three Types of Cognitive Load in Teaching
Intrinsic load refers to the natural difficulty of the material being learned. Some topics are inherently more complex than others. For example, students encountering calculus for the first time will struggle more than those with previous exposure. While intrinsic load can’t be eliminated, educators can scaffold learning experiences to make complex topics more digestible.
Extraneous load is the unnecessary mental effort caused by distractions or poor course design. Think about a PowerPoint overloaded with text, distracting animations, or unclear instructions. These elements waste cognitive energy and make learning harder than it needs to be. Effective teaching aims to reduce extraneous load so students can focus on the content that truly matters.
Germane load is the productive cognitive effort that helps students integrate new knowledge. This is where ChatGPT can be a powerful learning tool. As Mah and Levine (2023) emphasize, it’s nearly impossible for faculty to gauge students' prior knowledge and misconceptions. Traditional differentiated instruction can be time-consuming and ineffective if educators overestimate or underestimate student understanding.
How AI Can Support Learning
This is where AI revolutionizes education. ChatGPT allows students to customize their learning experience, adapting instruction to their individual needs in previously impossible ways.
For students struggling with intrinsic load, AI can break down complex topics, generate real-world examples, or provide step-by-step explanations tailored to their level of understanding. Imagine a student new to physics asking ChatGPT to explain Newton’s laws using sports examples while another student with prior knowledge asks for a deeper dive into real-world applications. ChatGPT can adjust to both learners, offering the kind of personalized scaffolding that traditional instruction often lacks.
Another significant advantage is reducing extraneous load. Instead of struggling through dense readings, poorly designed slides, or vague instructions, students can use AI to summarize key points, extract big ideas, and clarify complex concepts. If students feel overwhelmed by a research article, they can ask ChatGPT to highlight and break down the main arguments, making it easier to focus on relevant content.
And then there’s germane load, the mental effort required to connect and apply new knowledge. This is where AI is genuinely revolutionary. ChatGPT can help students restructure information in ways that make sense, rephrase explanations in accessible language, and generate relatable analogies that are individualized to their needs. When students actively engage with AI, asking it to reword, refine, or expand ideas, they synthesize, question, and make deeper connections.
AI and the Zone of Proximal Development: Learning with a Thought Partner
The Zone of Proximal Development (ZPD), developed by psychologist Lev Vygotsky (1978), explains that learning happens best when there’s just the right amount of support, not too much or too little.
Example: Think about when a child is learning to ride a bike. At first, they might need training wheels or someone holding the seat. Eventually, they gain confidence, and that extra support is no longer needed. That’s ZPD in action: learners start with guidance and gradually gain independence.
This is exactly how AI can function in education. AI, like ChatGPT, guides learners through their thinking process, offering support when needed but stepping back as they develop understanding.
Testing the Thought-Prompting Approach: A Personal Reflection
Simply reading about Mah and Levine's approach wouldn't have provided the diverse examples and methods essential for mastering new content. To truly grasp the material, it's crucial to actively engage with it. Typically, I input a stream of thoughts into ChatGPT, which usually works well. However, this time, I aimed to emulate Mah and Levine's strategy by being more deliberate with my questions and guiding the AI's responses. Admittedly, part of this was because I knew I'd be sharing my prompts to my professor and wanted to appear more methodical. I was concerned that this approach might slow down my usual rapid-fire method of seeking answers.
To my surprise, being more thoughtful about how I wanted the bot to respond significantly enhanced my experience. What I anticipated to be a mundane exercise became a valuable reminder of how applying learning theories can improve communication with AI technology.
Below are excerpts from my ChatGPT prompting and adaptation session. In intentionally crafted prompts to explore various examples, arguments, and perspectives on machine learning—the technology that powers ChatGPT itself. By thoughtfully modifying my prompts, I guided the AI to respond to different levels of complexity, enhancing the learning experience. This approach demonstrates how tailoring prompts can effectively utilize ChatGPT as a dynamic educational tool. You'll see how I intentionally crafted my prompts to guide the AI's responses, leading to a more effective and engaging learning experience. I aimed to demonstrate how modifying your prompts can encourage ChatGPT to provide examples at varying levels of complexity—a crucial skill for effectively using ChatGPT as a learning tool. By tailoring your prompts thoughtfully, you can guide the AI to deliver responses that match your desired depth and intricacy, enhancing your educational experience.
So, I did what any educator would do—I asked ChatGPT, “What is machine learning?”
At first, it gave me a standard definition:
"Machine learning is a branch of artificial intelligence (AI) that enables computers to learn from data and make predictions or decisions without being explicitly programmed."
That made sense on a surface level, but it still felt too abstract. How would I explain this to faculty who don’t consider themselves tech-savvy?
I asked ChatGPT for a simpler explanation. It adjusted:
"Machine learning is like teaching a student to recognize patterns from examples rather than giving them strict rules to follow. Instead of programming a computer with step-by-step instructions, we show it many examples and let it figure out patterns independently."
That was better. But then I had another question: How does the machine actually "see" patterns? Isn’t AI just a bunch of 0s and 1s?
ChatGPT adapted again:
"Yes, computers only understand 0s and 1s, but they ‘see’ patterns by breaking things down into numbers. For example, a black-and-white image is stored as a grid of tiny squares (pixels), each assigned a number representing brightness. AI looks at these numbers to detect shapes and features—just like how you might recognize a cat by its ears or whiskers."
ChatGPT was adjusting its explanations in real-time, scaffolding new information based on what I already understood. This mirrors effective teaching methods, where instruction is adapted to meet learners at their level and build on prior knowledge.
This process aligns perfectly with Mah and Levine’s (2023) argument that AI can serve as an adaptive example machine, helping students engage deeply with learning material through iterative questioning and refinement. But beyond that, it also reflects two key learning theories:
Cognitive Load Theory (Sweller, 1988) – ChatGPT naturally reduced extraneous cognitive load by avoiding unnecessary complexity and presenting information in small, digestible chunks. Instead of overwhelming me with too much information at once, it broke things down step by step, helping me focus on what truly mattered.
Zone of Proximal Development (Vygotsky, 1978) – Rather than simply giving me answers, ChatGPT acted as a scaffold, providing just enough guidance to help me move from what I already knew to new concepts. This interactive support mirrors how teachers help students bridge the gap between what they can do alone and what they can achieve with assistance.
This experience reinforced what I already suspected: AI isn’t just a tool for content generation—it’s a thought partner that actively supports the learning process. When used strategically, AI can help scaffold learning, reduce cognitive overload, and encourage deeper engagement, making it a valuable asset rather than a threat in education.
Why Faculty Should Try AI for Themselves
AI isn’t going anywhere; ignoring it won’t stop students from using it. Rather than resisting its presence in the classroom, faculty should take the time to explore and experiment with it firsthand. ChatGPT isn’t just a tool for students. It’s a powerful resource that can enhance faculty workflows, spark creativity, and streamline course design. AI has also proven to be an invaluable asset in training faculty on compliance initiatives. These tools help me develop effective training materials and generate relevant examples for disciplines outside my area of expertise, allowing me to create more engaging and tailored learning experiences.
As an instructional designer and compliance specialist, I’ve seen firsthand how AI elevates my creativity and productivity in ways I never imagined. It allows me to brainstorm faster, organize ideas more effectively, and design instructional materials supporting diverse learners. So, what are you waiting for? Give it a try; you might be surprised by the doors it opens for your students and your teaching.
References
Mackh, B. M. (2021). Pivoting your instruction: A guide to student-centered, equity-focused learning design. Routledge.
Mah, C., & Levine, S. (2023). How to use ChatGPT as an example machine [Podcast 205]. Cult of Pedagogy. Retrieved from https://www.cultofpedagogy.com/chatgpt-example-machine/
Stanford Institute for Human-Centered AI. (2023, July 10). AI detectors are biased against non-native English writers. Stanford University. Retrieved from https://hai.stanford.edu/news/ai-detectors-biased-against-non-native-english-writers
Sweller, J. (1988). Cognitive load during problem solving: Effects on learning. Cognitive Science, 12(2), 257–285. https://doi.org/10.1207/s15516709cog1202_4
Vygotsky, L. S. (1978). Mind in society: The development of higher psychological processes. Harvard University Press.
Watermark Insights. (n.d.). The evolution of higher education. Watermark Insights. Retrieved February 2, 2025, from https://www.watermarkinsights.com/resources/blog/the-evolution-of-higher-education/
Comments