Post and discuss your response to the reflection question - What was one challenge and one success you had when using a LLM to support your work as an educator.
I was able to use the tools to get a new idea for a lesson plan. I did have trouble with the LLM not really understanding what I was asking, but clarifying and prompting again did help.
I was successfully able to generate a formative assessment and a project assessment. I felt sucessful in being able to continually prompt ChaptGPT until it was something I could use. I will be teaching a unit on AI. I think my biggest challenge is that I didn’t feel comfortable enough with the content. I didn’t feel I could judge the assessment because I didn’t have a good understanding of the material. I redid the activity again with a topic I felt I could judge the response of ChaptGPT and was very happy with the result.
I asked ChatGPT to create a lesson plan for a professional development session on the Nevada School Performance Framework and provide opportunities for participants to reflect on how they can contribute to the growth of scores on the NSPF for their school. I then was able to successfully have ChatGPT create a handout for teachers to record their own school’s data and record actionable steps they will take to support the school’s work in each of the 5 areas. A challenge I had was getting ChatGPT to come up with a description and scoring criteria for the indicators and measures that was accurate for the current scoring system that is being used.
I had ChatGPT write a formative assessment based on some digital citizenship lesson plans from a website. It gave me some good questions, but it didn’t quite cover every topic. I then prompted it to add a few questions about the topic it had missed and it was able to update the assessment. I then asked it to round out the assessment so that the total number of points equaled 30, which it was able to do and assigned each questions points and added a few questions. Overall, I’m pleasantly surprised at the results.
I had ChatGPT construct an email to a parent regarding a reluctant reader, much like in the video example, but I narrowed the criteria further and added both positives about the student and suggestions for ways the parents could support the student at home. The final message was clear and professional. It didn’t have many specifics about the child, of course, but those could be manually added and aren’t always needed. The message was successful overall, but for me, it would have probably been faster to write the email than to have AI write it for me, especially after adding refining criteria. Every teacher and situation is different though.
I asked ChatGPT to help write an email to the parent of a disruptive student. I tried adding more details about the type of disruption, but I struggled with this part. It seemed when I added a specific detail, it kind of replaced another detail I wanted to keep. I have to figure out how to be specific enough in my prompting without sacrificing things that had already been good in the original email. I did like the way the AI worded things–I found that this was better than I often do myself, and it had a more pleasant tone overall.
I chose to have it create a formative assessment. One challenge is that it had very specific items on it that may or may not have even been taught when I taught the unit. Which just means I would have to just make sure I started with the formative assessment on AI, then made sure to include all information on those specific items to make sure the kids were assessed fairly. But the entire thing was a success - it generating multiple choice questions and even going into detail about difficulties on certain questions the kids might have is really cool to know. I can easily plan to fill in those knowledge gaps this way.
I used Magic School first to create a lesson plan and the challenge was it was not detailed enough. I used the same criteria of creating an engineering notebook lesson plan and decided to make an engineering notebook standards assessment. The success was the assessment was the exact information we covered in class and I plan to use it next school year.
The option I chose was to use ChatGPT for writing a parent email. One success I have is learning how to “trick” the AI into providing better responses by including empathetic language (like “please,” “thanks,” and “do your best because this is really important to me”), which might be a result of human bias. I also asked ChatGPT to rate itself and list potential improvements. Overall, the email it generated was decent!
One challenge is that the context of the email I asked ChatGPT to write was to address student behavior in a class I had today. Obviously, ChatGPT has no context of myself as an educator, my class, the student in question, or the student’s family. Thus, the email reads pretty coldly and generically. Personalizing the message through the potential improvement areas and personalization definitely makes a better message!