Teacher Tech blog with Alice Keeler

Paperless Is Not a Pedagogy

Alice Keeler

AI is NOT Magic – The Need for Expertise

As much as we wish AI would do it for us, remember that AI is not magic. To get quality output you have to be an expert.
AI is NOT Magic – The Need for Expertise

“Can’t AI just do that for me?” This question, echoing in teacher lounges and online forums, reflects a common misconception about the role of artificial intelligence (AI) in education. It’s tempting to believe that AI can magically generate lesson plans, assessments, or even entire curriculums with the click of a button. But the reality is far more nuanced. AI is NOT magic; it can’t read your mind. In fact, using generative AI can be frustrating at times, as it requires continuous refinement and “arguing” with the output to truly create educational magic.

AI is Not Magic - It Helped Me Create this Blog Post

Did I use generative AI to help me with this blog post? Yes. Did it write this blog post? No. Why? Because AI is not magic. Generative AI is great at helping me use semicolons correctly. However, it is not great at understanding who I am and what knowledge I am trying to get across. I used Google Gemini and ChatGPT to help me write this blog post. See at the bottom for screenshots of my prompts.

I wrote the above paragraph (and this one) with zero help from AI. Notice from the screenshot that I copied and pasted it into a tool to improve my grammar. The result was more words or not enough and ultimately I was happy enough with what I originally had. 

You Must Be a Teaching Expert

A trained teacher brings a wealth of nuanced understanding to lesson planning, feedback, and assessment that AI models struggle to replicate. They possess the ability to gauge students’ emotional states and engagement levels, tailoring instruction and feedback accordingly. Teachers draw upon years of experience and pedagogical knowledge to create lessons that are not only content-rich but also developmentally appropriate and culturally responsive. In providing feedback, teachers go beyond identifying right or wrong answers, offering personalized guidance that fosters critical thinking and growth mindset. Similarly, in assessment, teachers can design evaluations that truly measure student understanding and identify areas for improvement, taking into account individual learning styles and needs. AI models, while capable of analyzing data and generating content, lack the emotional intelligence, cultural sensitivity, and nuanced understanding of student learning that are essential for effective teaching.

AI is NOT magic; it requires the wisdom, experience, and expertise of seasoned educators to truly make a difference in the classroom.

The Illusion of AI Infallibility: Trained on Flawed Content

Adding another layer of complexity, it’s crucial to remember that AI is trained on vast amounts of data from the internet—a treasure trove of information, but also a breeding ground for misinformation and bias. From outdated pedagogical approaches to inaccurate content, the data AI learns from isn’t always a model of educational excellence. This means that AI-generated content can sometimes perpetuate these flaws, leading to lesson plans that are pedagogically unsound or assessments that don’t accurately measure student understanding. As a result, blindly trusting AI-generated content without critical evaluation can be detrimental to student learning.

AI is not magic and understanding “The promises and perils of AI in education” is essential. Ken Shelton and Dee Lanier have created a book to assist with getting past the magic to allow for better use of generative AI. 

AI is a Collaborator - Not a Magic Wand

Think of AI as a collaborator, not a replacement. It’s a tool that can augment your expertise, but it can’t replicate it. The quality of the output depends heavily on the quality of the input. This means crafting detailed prompts that go far beyond simple requests like “create a lesson plan on photosynthesis.” Instead, think of your prompts as a window into your teaching expertise. Share your deep understanding of the subject matter, drawing upon your years of experience and knowledge. Explain the nuances of the topic, highlight potential misconceptions students might have,

Is Using AI Even Worth It?

A reading specialist, fresh from an intensive training on a new phonics-based program, reached out to me recently, her voice laced with frustration. The program was robust, promising impressive results, but the workload was immense. Individualized lesson plans for each student, meticulously aligned with the program’s intricate scope and sequence, felt like an insurmountable mountain. She had hoped AI could be her magic wand, conjuring up these detailed plans with a few keystrokes. The reality, however, was far from enchanting.

Time Saving But Lackluster

The AI-generated plans, while technically aligned with the program, lacked the nuance and depth she desired. They were generic, cookie-cutter outlines that failed to capture the unique needs and learning styles of her diverse students. The magic she had hoped for felt more like a mirage, leaving her disheartened and questioning whether the effort to integrate AI was worth it.

This experience is not unique. Many educators yearn for AI to be a shortcut, a time-saving miracle. But the truth is, AI is not a mind reader. It cannot understand the subtleties of a specific program or the individual needs of each student without explicit guidance.

The magic, it turns out, lies not in the AI itself, but in the expertise of the teacher.

Teaching is Time Consuming and Requires Expertise

By investing time and effort into crafting detailed prompts that capture the essence of the program, the specific needs of each student, and the desired learning outcomes, teachers can transform AI from a generic content generator into a personalized learning tool. This process requires patience, a deep understanding of the curriculum/program/teaching practices, and a willingness to experiment and iterate. But the payoff can be significant: lesson plans that are not only aligned with the program’s scope and sequence but also tailored to the individual needs of each student, fostering deeper engagement and ultimately, better learning outcomes.

Debunking Myths: The Learning Styles Example

One significant risk of using AI without sufficient expertise is falling prey to misconceptions. For instance, the myth of learning styles has been debunked, yet the internet is rife with this false information. AI might produce outputs based on these incorrect patterns simply because they are mentioned frequently. This example highlights why teachers need to be highly trained experts and not simply rely on AI. It is crucial to recognize and rectify such issues to prevent the spread of misinformation and ensure that students receive accurate and valuable knowledge.

Learn More about Google Docs from Teacher Tech

How Long is Your Prompt?

If your prompts are mere sentence fragments, the results you receive from AI might leave you wanting. To truly harness the power of AI in education, consider crafting prompts that are nearly a page long. Think of it as a brain dump where you pour out everything you know about the topic, your students, your values, and your educational philosophy.

Messy Think Everything You Know About the Topic and Teaching

Let your thoughts flow freely, capturing the nuances of your teaching practice. Describe your students’ diverse interests, and challenges. Articulate the specific skills and knowledge you want them to acquire, as well as the values and soft skills you hope to instill. Don’t be afraid to get messy and include seemingly unrelated details or tangents; these can spark unexpected connections and lead to innovative ideas.

  • Curriculum Standards and Goals
  • Content Knowledge
  • Pedagogical Approaches
  • Resources and Materials
  • Student Needs
  • Anticipated Challenges
  • Classroom Dynamics
  • Values and Skills
  • Differentiation Strategies
  • Examples and Analogies
  • Assessment Methods

AI as Your Collaborator Reflects YOUR TRAINING and Not Patterns from the Internet

By providing AI with this wealth of information, you’re not just asking it to generate a generic lesson plan. You’re inviting it to collaborate with you, to synthesize your knowledge and insights into a cohesive and effective learning experience. Think of AI as your thought partner, helping you to organize your ideas, identify gaps in your thinking, and generate creative solutions.

This process of crafting lengthy, detailed prompts may seem daunting at first, but it’s a worthwhile investment. It forces you to reflect deeply on your teaching practice, clarify your goals, and consider multiple perspectives. In the end, the resulting AI-generated output is likely to be far more nuanced, relevant, and impactful than a generic lesson plan produced from a few keywords.

Books by Alice Keeler

Available on Amazon or in bulk order.

Continuous Professional Development is the Magic

Being a teaching expert means committing to continuous professional development. The field of education is ever-evolving, and staying updated with the latest research, methodologies, and technological advancements is crucial. By continuously enhancing your knowledge and skills, you ensure that you can effectively integrate AI into your teaching practice. This ongoing learning process helps you stay ahead and make informed decisions about using AI tools.

Argue Argue Argue - The Expert is YOU Not the AI

The most effective way to use generative AI tools is to engage in a dialogue, a back-and-forth exchange where you challenge, refine, and ultimately shape the output. Don’t be afraid to argue with the AI! Remember, you are the expert, the one with the deep understanding of your students, your curriculum, and your pedagogical goals.

AI-generated content should never be taken at face value. While AI can process and analyze vast amounts of data, it lacks the contextual awareness and critical thinking skills that you, as an educator, possess. More importantly, AI is often wrong. If you are not finding errors in its outputs, it might indicate that you need more expertise training. This is why it is essential to argue with the AI’s outputs. By questioning and critically evaluating what the AI produces, you ensure that the content aligns with educational best practices and the specific needs of your students.

Ensuring High-Quality Outputs

Arguing with AI means continually refining and improving the prompts and outputs. Here’s how you can do this effectively:

  1. Scrutinize for Accuracy: Always verify the facts and information provided by the AI. Cross-check with reliable sources and your own expertise to ensure the accuracy of the content.

  2. Challenge Assumptions: AI can sometimes perpetuate outdated or incorrect assumptions. Challenge these assumptions by applying your knowledge and understanding of current educational research and best practices.

  3. Refine and Edit: AI outputs often require significant refinement and editing. Don’t hesitate to rephrase, adjust, or completely overhaul the generated content to better suit your objectives and standards.

  4. Address Biases: AI can inadvertently incorporate biases present in its training data. Be vigilant in identifying and correcting any biased language or perspectives to ensure fairness and inclusivity in your teaching materials.


Use GPT's and Gems

Below are the links to the lengthy conversations I had with both Google Gemini and ChatGPT. What you don’t see in the transcripts is that I have pre-trained all of my prompts to ensure the outputs closely align with the values of Teacher Tech.

I subscribe to both ChatGPT and Google Gemini Advanced. The paid versions not only provide better output but also do not use your data to train their models. 

Another advantage of the paid models is the ability to create custom chats that you can pre-load with your teaching pedagogy. These are easy to create.

GPTs for ChatGPT and Gems for Google Gemini allow you to pre-write a set of directions. I suggest making them as detailed as possible, including everything you know and value about teaching. For example, when I type in a generic topic such as “quadratic formula,” I get a fairly decent starter for my lesson plan that aligns with my teaching values. This is where AI really starts to save me time.

This Blog Post Took Me 3 Hours to Write

Yes, I used AI and yet there was still no magic wand to just write it for me. Hardly feels like magic when it takes so long. Might have taken me longer than just doing it myself. While I had to put significant effort into giving AI my content and argue with it to beat it into submission, I feel the result was better than manually writing the post from scratch. This post is far from AI generated content though. It is fully me. My ideas, my writing, my vetting, my editing, and my values. And some paragraphs are 100% AI free.

Highlights of My Conversations

I used both platforms. Sometimes I pasted what one platform said into the other to try to improve the output. I took the same prompt and copied and pasted into each to see which one would do better. What you can see from these conversations is how much input I put into the conversation. What you do not see is that I used Google Gems that pre-slotted directions that align to my Teacher Tech values. 

My initial prompt for Google Gemini is so small because I have created a Gem that outlines how I want to write my blog posts. It already knows I am writing a blog post for Teacher Tech. 

Here is My Content... Ugly

I actually largely ignored the first draft that Google Gemini spit out. Not because it wasn’t good, but because there was clearly not enough of my own input. What you can see for my follow up prompt is that I basically wrote everything I wanted to say in the blog post but without nicely formatting it. I then expect that moving forward that the generative AI is helping me clarify my points, not be an expert on the topic. I also pasted the exact same prompt into ChatGPT. 

Over to ChatGPT

My next prompt into ChatGPT was asking it to improve my paragraph for grammar and active voice. 

Keep Arguing and Refining the Prompt

From the conversation you will see a lot of me pushing back with the AI. Explaining that it is leaving out information or that some of the information is misleading. 

Building Each Section

Rather than trying to get generative AI to spit out a perfect blog version, I work with it to craft the wording for each section. Notice in this excerpt I am again letting ChatGPT know what I want it to say rather than relying on it create content for me.

Modifying the Writing Style

I want to emphasize certain points. I asked Google Gemini to help me reorganize the emphasis in paragraphs. 

Using AI in the Classroom

I have and do use Google Gemini and ChatGPT to help me write lesson plans. However, if I just ask for a rubric it spits out mathematically flawed scoring guides. Are the majority of the rubrics on the internet exemplars? Certainly not, and the poor quality ones helped to train the models. What do many lesson plans say? Talk at the students and have them fill out a worksheet. That is not my teaching style, yet low quality lesson plans are part of the training models and thus they influence the output generative AI spits out. 

Since AI models are trained on not only high quality exemplars there is a high probability that misinformation, bias, and simply low quality show up in the teaching materials. 

The only way to combat this, is for you, as a teacher, to be highly trained. That you have enough expertise that you can identify what is factual, what is good for your students, and what makes good education. Use AI, but load up the chat with as much about high quality teaching as you can. Realize that the models were trained on low quality materials and it is up to you to critically analyze and argue with it. This is no magic wand because AI is not magic. It’s as good as it’s inputs and you need to be the most valuable input and this takes time. Be prepared to spend time working with AI rather than trusting it. You’re the expert. 

© 2024 All Rights Reserved.