I’d be pleased to entertain ideas about how to most effectively use Claude in generating the performance expectation language, or in other aspects of PE Maker. For example, if you have an idea for a more effective prompt, please let me know. Or, maybe it’s necessary to take a step back. How would we measure, or at least recognize, a better result? What is “better” PE language? Let me know your thoughts.
I have had fairly good success using Claude for my NGSS assessment development and think it is a good choice for this task. Here are a few thoughts:
-How similar in language is desirable for PEs – i.e. should there be more or less variability? (Develop models to describe, Develop and use a model to describe, Develop a model that)
-I’m a fan of progression charts that support teacher awareness of the concepts that come before and after a PE, and I would advocate for including a way to visualize and recognize the interconnectedness of concepts. Do you think some type of standards progress mapping tool (maybe similar to McGraw Hill’s ALEKs math) would be useful and practical?
-Are any current clarification statements or assessment boundaries part of the prompt guardrails so the PE AI tool does not go beyond the grade level intent?
Happy to discuss further!
Becky:
-I would probably lean toward less variability. Less variability, more portability.
-I too value representations of connections. I need to figure out how to best approach that in StandardsView.
-In a word, No! What you can see on the Prompt and Examples page is the entire instruction to Claude. Off the top of my head…at the moment, Claude’s only getting general instructions about how to build a PE–any PE–every PE. Sending instructions that are specific to the PE being designed is a different matter. You’d have to identify the relevant info to send. I’ll think on this. It would be great to find that someone has already collected the existing clarification statements and assessment boundaries into a useful whole.