3 things AI and robotization should already have taken over in school administration

In the area of school digitalization and the departments “administration” and “pedagogical administration” there are a lot of things that should already have been automated, with or without AI. Here are simply a few examples of things I think we really shouldn’t be spending time on anymore, and that I believe some smart algorithms could do much better — freeing up time for other things.

1. Scheduling

Scheduling is an area I don’t understand how isn’t already automated. At the start of the scheduling work we enter a lot of data anyway; number of seats in rooms, which facilities can be used for what, which teachers have reduced hours and which days they can’t be scheduled, and much much more.

You just keep adding to that: minimum break minutes — depending on whether it’s a theoretical subject or, for example, PE (to have time to shower), preferred lesson length, length of the school day, prioritizing that certain classes should have certain classrooms, mixing theoretical and practical subjects during the day and other things. With all that data it should be possible to train an AI to produce the best possible timetable based on each school’s conditions. If you have a sufficiently large dataset of what constitutes a “good timetable” that analysis should make the AI even better. There should reasonably be research on how to structure a school day to achieve the best effect on pupils’ learning — in short, what a “good timetable” is. Often the principal or vice principal makes the timetable. It’s not easy and it takes many weeks, often at times when the principal should be focusing on a new school year instead. Also, no one is ever satisfied with their timetable, so a bonus effect is that we can blame the algorithm. ;)

Sidequest | IoT - connected rooms and timetables

Another part of scheduling could be that if we had sensors and similar in every room, the timetable could automatically suggest and make temporary changes during extreme weather such as a heatwave — for example before a national exam — when temperatures in some classrooms can become far too high. Or free up rooms that aren’t actually being used. The data each room collects could be analyzed to produce forecasts. I’m tinkering with a prototype around this, which you can read about here.

After relatively little [!] googling I haven’t found any companies offering automatic scheduling or research on what a good timetable is. Please share tips.

2. Analysis work

In most school platforms today we collect enormous amounts of data about students and their results. This can be, for example, subject assessments in grades 1–5, national tests in grades 3, 6 and 9, UIM in grades 2, 5 and 8, grades in grades 6–9, user surveys, screenings, extra adaptations, absence and probably a few more things. If we also bring in external items like SALSA and information about staff, such as absence — that is, whether students have had lots of substitutes, qualified teachers, number of students per teacher and other things, we should have more than enough data for analysis. We should be able to follow pupils in real time from pre-school and onwards. Here AI should be able to work through all the data that already exists to find common denominators regarding warning signs, but of course also positive trends. All to be able to support where needed and reinforce what’s working. By doing this analogously in miniature, I’ve identified a number of exciting things at our school. What data do you use at your school when you do your analyses — and how often? This year’s grades on the last days of June? Instead of spending a lot of time making scant summaries, analyses and trend reports should be available throughout the school year so we can spend time actually discussing them and coming up with solutions. This could become a fantastic tool for school leadership and teachers to, for example, know how to allocate the school’s resources and where we need to implement support measures (for both students and staff) — and will need to implement support measures, before it’s too late. It could also become a fantastically unpleasant tool, but we can take that another time. The alternative is that we stop collecting data altogether. If we don’t use the data, we shouldn’t collect it.

3. Chatbots

Simple administrative chatbots could take some of the questions that come into school offices. For example by handing out various forms, like leave request forms, contact details and other things. The nice thing is that the bots can be trained to understand the intent of someone who doesn’t know they should search for “leave request form” on the school’s website, but searches for “ledig” or “semester”. A bot can understand what is meant. The more people use the bots, the better they can be trained. Of course these bots can be plugged into all the channels the school has, either via search or APIs. At Årstaskolan we’ve had a bot on the school’s Facebook page for a couple of years. Unfortunately I haven’t had much time to work on it, but it’s there for you to test. In the video below you can see a short conversation I have with it.

[KGVID]https://objects.dc-sto1.glesys.net/aged-recipe/2019/09/chatbot.mp4[/KGVID]

What else should already be automated?

Feel free to comment in any forum you like or here on the site. Personally I also think the following things could be automated; * Automatisk vikariehantering skulle ta bort ett stressmoment från administrationen. Borde inte heller vara särskilt svårt att få ihop.

• En grundbudget borde likt schemaläggning, kunna göras automatiskt.
• Ansikts- / eller annan biometrisk avkänning och rapportering i närvarosystem, även om det är juridiskt klurigt.