NN4DA in the News

Artificial Intelligence and Charter School Authorizing

Implications and Applications for Charter School Authorizing by Districts

 

The NN4DA is launching an initiative to facilitate district authorizers’ consideration and exploration of Artificial Intelligence’s (AI’s) implications and applications for authorizer work. I invite you to join in that work. This update includes information on NN4DA’s proposed work on Artificial Intelligence (AI) and a few resources on AI that may interest charter school authorizers beginning to explore AI.

As part of that effort, we are conducting a quick survey of authorizers’ AI use. Please take a moment and complete this NN4DA AI Authorizer Survey (even if you don’t currently use any AI in your work). We want to learn about how much is used, for what purposes, what district policies are in place, and what kind of help people want in the future.

To advance the discussion, we will meet on the topic of AI, in conjunction with an upcoming meeting of the Florida Association of Charter School Authorizers (FACSA). Some of us plan to come a day early for an NN4DA in-person symposium on AI in authorizing, on Thursday, May 1. The FACSA meeting is Friday, May 2, in Fort Myers, Florida. FACSA is delighted to have people stay over and join their meeting. If you are interested in joining, please contact me directly at Alex.Medler@NN4DA.org. Before you ask, the NN4DA doesn’t have funding to pay for travel.

If you want to read more about AI, this NN4DA AI Resource List provides a list of reading. For starters, I recommend a fascinating blog, Think Forward: Learning with AI, from CRPE’s Robin Lake. For a skeptical take, I would start with Benjamin Riley’s Cognitive Resonance blog.

On a personal note, I have to disclose that I am perhaps the worst cynic you know when it comes to the application of technology to student instruction. But, if we leave AI’s application to student instruction aside, when it comes to the work that charter authorizers do, my views are more positive. I hope others that might share my ingrained skepticism, as well as those who love all new tech, and everybody in between, feel welcome to join this effort. We need each other for this consequential work.

I recently did a little experiment. In my role as school board member, I am going through a charter application review. I tried to stick to my role as a board member, and left it up to the district staff, who worked with CACSA, to conduct a thorough charter application review, with all the trimmings. I then received a packet containing about 300 pages of material.

The folder included a 125-page application, and a longer revised application, an 18-page review by district staff and a 40-page review by a District Accountability Committee (DAC). Both of these reviews used a high-quality application rubric, rigorously and systematically reviewing all application sections. It also included capacity interview questions from a long deck and the written response from the applicants after the capacity interview in response to follow-up questions that the applicant didn’t have time to address during the interview.

What Happened When I Test Drove AI with Authorizing Tasks?

Despite my own bias against ed/tech, after working with few tools to assist with common authorizing tasks, using real charter application material, I’m confident that today’s commercially available AI tools can improve or enhance our ability to perform some of the analytical tasks that are required of authorizers now. As a profession, we may have already arrived at the point where we should explore and consider applying AI to some of the professional tasks performed throughout the life-cycle of a charter school.

What did I learn about AI’s potential?

  • AI is a lot faster than me.
  • AI organizes its notes better than me.
  • Often, it wasn’t wrong — or at least it wasn’t different from me.
  • We agreed about areas needing more investigation in the application.
  • AI could turn this analysis into tools for sharing with other audiences quickly and easily, and these products didn’t suck.

On the less positive side:

  • It was really bad at independently appraising the quality of answers.

What do I think after this exercise (which was admittedly very limited and crude)

Charter authorizers’ work — what technologists would call their “use case” — involves several activities that Large Language Models (LLMs), like ChatGPT and NotebookLM, are pretty good at performing today. With the power of these tools off-the-shelf, it is provocative to think about what customized applications designed specifically for our purposes could do in the future.

Authorizers’ analytical functions include:

  • Collecting and organizing a large and ever-growing body of evidence;
  • Tracking all the relevant places where, in a huge volume of information, a particular issue is addressed in text;
  • Summarizing a large volume of data, in different forms designed for specific audiences;
  • Synthesizing information from many sources;
  • Comparing and contrasting different stakeholders’ input regarding a single case; and
  • Creating emails, memos, documents, and presentations to communicate findings and analysis.

Other important work we do is different, and not likely to be suitable for AI at this time. The expert judgement and appraisal of capacity or the quality of plans is a task I would not hand-off. That said, the curve of apparent improvement among AI tools is steep indeed. Who knows what this will look like in two years.

My bottom line, at a purely functional level, authorizers should start exploring AI. As the AI resources linked at the start of this blog attest, there are tons of issues to address during this conversation that I haven’t touched on here. Feeding LMs proprietary or non-public data, application quality inflation from applicants using AIs to write applications, policy obstacles on use, bias, and time wasted fact-checking the failures and hallucinations. These are all real issues to tackle. But tackle them we must. I hope you will join me in that exploration.

Look for more opportunities to join this work in the coming months. And please reach out with any feedback, questions, or suggestions.

Alex

P.S. don’t forget to take two minutes and complete the survey!

Scroll to Top