How I use Artificial Intelligence for planning
We need to get off the hamster wheel
For the reasonable price of just $5 AUD per month, you can be a full subscriber to Filling the Pail with complete access to the extensive archive on a range of topics, as well as access to the unabridged version of my weekly Curios posts.
Over on Twitter/X, Adam Boxer shared a post from a user called ‘leah’. The post included a photo of a homework assignment that looks like it may have been generated using AI.
Some will be of the view that teachers should not use AI at all when planning lessons. My teenage daughters are strongly opposed to any AI use, convinced it is contributing to the all-encompassing climate emergency that will rob them of their futures—an interesting perspective to explore in a different post.
Nevertheless, it’s not hard to understand why teachers might use AI. In Australia and the UK, for example, there is often the expectation that teachers should plan all their own lessons. Sometimes this planning is supported by a textbook but often it is not—one maths department I oversaw in the UK decided textbooks were the enemy and banned them. Which is obviously bonkers.
The philosophy behind the hamster wheel approach to planning was exposed a few years ago in Australia when the Grattan Institute issued a report about schools—including mine—that engage in shared lesson planning. Surely, the authors reasoned, giving lesson plans to teachers would save them a lot of time.
In response, academics wrote articles like this one on the AARE blog, outlining their concerns.
“Teachers delivering content via a pre-prepared script or lesson might seem easier and simpler, but it remains difficult to see who benefits from a lifeless and unthinking teacher delivering someone else’s content. The key to teaching – and learning – lies in the human relationships between teachers and students. Those human relationships allow for careful contextualisation and design.”
Lifeless and unthinking? Please, tell us what you really think.
Even if you believe that lesson plans need to be somehow adapted to the individuals in the class you teach—something I am sceptical about—why does giving a teacher a plan prevent this? At least teachers then have a starting point and if they want to jazz it up or personalise it, they can. This is not about logic, it is about an ideology that is alive and well in the university education faculties whose members do not have to live with the consequences.
The truth is that nobody can do all the planning from scratch that this ideology requires. Instead, teachers resort to Googling worksheets or even paying other teachers for resources. Since the advent of AI, teachers have had access to a bespoke planning tool and of course they are going to use it.
I am teaching Year 12 physics this year and I am making use of AI. However, I don’t have to plan the whole course. I am starting with the 2025 version and have been asked to simply update one of the units. I am therefore not on the hamster wheel and instead of using AI as a survival tactic, I can use it more strategically. I can make minimal changes if I choose because the 2025 version is good enough to roll out again.
Nevertheless, I am working on a few tweaks. For instance, I generated some transparent images of sine waves and cosine waves that are handy for updating some of the diagrams.
I also generated 20 questions that involve calculating magnetic flux, magnetic field strength and area. There’s a planning concept known as ‘Shed Loads Of Practice’ or ‘SLOP’ that teachers have been blogging about for years. Unfortunately, I’ve found that if I use this term at work, some people assume the ‘S’ stands for something else, so I’ve taken to calling it ‘fluency practice’. Worth a post in its own right, the basic idea is to practice past the point of success—a form of overlearning. If anything, the physics unit I am working on could use a little more fluency practice and so the ability of AI to quickly generate lots of similar questions is helpful.
Although quick to generate, copying and pasting these questions with their equations into a PowerPoint slide is far more fiddly than it should be and definitely not something I could do if I was simply racing to prepare lessons for the next day.
Yes, I do use PowerPoint, I know some people hate it and I don’t care. It’s just a medium—one that can be used well or badly. We project our slides onto a whiteboard so we can write on them and that works for us. You do what works for you: visualisers, chalk, a stick in the sand, whatever. I couldn’t care less.
And much the same can be said of AI. It’s just a tool and one that can be used for good or evil, like PowerPoint or like a kitchen knife. And as with all tools, it has strengths and weaknesses.
AI currently has two main limitations. Firstly, it is inaccurate. You have to know enough to spot its errors. If not, you will teach falsehoods and that is bad. Unfortunately, I assume the hamster wheel of planning means that many students are now being taught factual errors.
Secondly, AI is horrible with images and diagrams. If you want to use image generation then you are in for a lot of faffing around and you might not even get there. The sine and cosine diagrams I generated each took about three rounds of prompting and I am sure there are art packages that would have generated them more quickly.
But that’s also a strength of AI. I don’t know what those art packages are and if I downloaded one, I would not know how to use it. I would probably ask AI. With AI, I can interact in plain English and so the on-ramp is much easier. This ease of use and the ability to interact in English rather than through non-intuitive menus means AI is not going anywhere.
Love it or hate it, teachers will use AI. If we don’t want them using it to create content of dubious value, then I suggest we need to give them high-quality lesson plans they can teach from directly or tinker with to their heart’s content. That way, when teachers use AI, it will be for slowly and deliberately making improvements or trialling innovations rather than to quickly produce something for tomorrow’s class that is stupid and full of errors.
Very well said. There was a time when all curriculum was centrally designed and timetabled in Victoria. At 9am Friday, eg, every Yr X student in the State would be doing English. Well-designed & in the right hands (therein lies the massive rub!) this, again, may prove a remedy.
Have you tried asking the ai for the best approach to generate a particular diagram.
Using tools seems to be a good use of AI.
They do better than a search engine in terms of presenting options and tradeoffs and can go further and generate input to online tools like wolfram alpha or other free tools.