I know this is for 527 but my reflection begins with LRNT526.  My experience with generative AI was non-existent, and I hadn’t considered it a topic worthy of serious attention. Up to that point I assumed my ARP would focus on K-12 education, with the specifics to be determined later.

After reading Teacherbot: Interventions in Automated Teaching (Bayne, 2015), I initially wanted to explore the possibility of robots replacing teachers. Intrigued, I delved deep into academic research, which revealed the enormous potential of machine learning in the classroom.  However, I quickly realized that AI, despite its capabilities, still falls short in replicating the complex interpersonal dynamics that even the least experienced teachers manage effortlessly. Unless we enter a full-blown Orwellian scenario, there will always be a need for human presence in the classroom as we know it.  And that was ultimately the key takeaway from all that research.

However, the Demystifying AI resource provided by our instructors opened my eyes to some practical applications of this technology.  I created fictional characters, images, and animated them.  I even created an entire art gallery using one prompt and a suite of different filters.  Really neat stuff that teachers can use today, but the resources on the project site were all grey lit and not sufficient to write a paper on.

As a professor at the Wharton School, University of Pennsylvania I was Inspired by Ethan Mollick’s research and his suggestion to start using Large Language Models (LLMs) and learn about them: their nature, information sources, and true capabilities. I explored complex issues surrounding copyright, trademarks, and patents, particularly focusing on the regulatory questions surrounding AI-generated content. Who owns the rights to AI-generated patents? The company that owns the computer making the AI request, or the person who prompted it? These questions still require formal regulation.  But this is a global technology, and the rules can be different in one place to another.  

Despite some limitations in function and questions on appropriate use, these AI systems are capable of remarkable feats. Sure they can enhance writing with guides and narratives, but they really excel at coding.

So upon starting LRNT527, I wanted to create something that would help teachers use AI more effectively. I didn’t want to create a slideshow on how to prompt AI, I wanted to build something tangible. With my very limited coding experience (just one MOOC on Python), I had never imagined building something like this prototype, but I discovered in 526 that AI could.

I basically built the prototype in 3 weeks. The first week was getting the templates made, the second spent on making them functional, and a third to get the project online. Because of my inexperience coding, it required many revisions on the part of the AI to output what I wanted. As a result, my coding skills improved. I was able to take snipets of code instead of requiring Claude to output “a full and complete file” that was never quite right. I also accomplished my goal of building a working prototype by the end of the course.

This project is currently a proof of concept, but it’s now online for you to try.  I envision developing it into a viable product for my ARP. Based on feedback, I would have improve the graphical user interface, Implement a login system for saving work to user accounts, and add features that enhance the portal’s value:

   – Creating an archive or library of generated activities and unit plans

  – Implementing a rating system for teachers to evaluate resources

   – Developing a system for sending activities and assignments directly to students

Building this prototype has been an incredible learning experience, showcasing the potential of AI in education and my own capacity for growth and innovation. I’m excited to continue this journey, expanding on this foundation to create a tool that can truly help educators and enhance the learning experience for students.