No, I think talking to AI this way forces teachers to think much harder about the learning process and student misconceptions. This will accelerate their progress as teachers. I love the way you keep returning to test AI capabilities.
There are two things that are on my mind after reading this.
1) we have to test AI created resources long term to see whether it works - my instinct is telling me that there will be certain repetitiveness of the design.
2) as a teacher I have to read and consider many incorrect solutions. Using AI expects me to fact check again. So I will be reading a material that is incorrect. It may be actually easier to create a resource. Not saving the time. It is more about the mind being exhausted from constant task switching and focusing on creating a task (without AI) may allow my mind to focus in a different (undisrupted) way.
Hi I read this article with interest. I am not a Maths teacher in fact I teach Religious Studies Ethics and Philosophy. I liked the Smith’s rubric in the article and asked Gemini to use the article and the rubric to create a worksheet that I could use as an end of topic task for one of the topics in GCSE RS Ethics. I was pretty impressed with the results as a first attempt.
I have been playing around with this as a primary school maths teacher/leader and I have to say I think it is fantastically simple. The opportunity to demonstrate within the model as well create something more child friendly is also incredibly easy to add. I love how simple it is to change and vary the outcome to better match what I need. For example, I made a rounding to 10, 100 and 1000 sheet. I asked for a snazzier title, more supports (like numberlines), more questions for question 2 and larger font. Done. In seconds. But, the beauty is each new iteration can use the references and refinements I have already made. Very much something I will be looking to explore more in the future.
What I want to know is why can't the pro versions of AI solve the "produce nice-looking mathematical expressions and images natively" problem yet? And don't get me started on diagrams. Not even ChatGPT5 can get those right. I've found that by the time I correct all the errors that AI-produced maths questions are riddled with, it takes about the same amount of time as creating them from scratch myself.
I asked Chat GPT to make me a graphing inequalities worksheet the other day, with graphing axes, and it gave me what appeared to be text based axes, rather than images. I assume it's just a limitation of the platform of which I was not aware since I am new to the use of AI. I ended up just making a worksheet but it was nice to have the list of problems generated by AI, even though I could probably just have thought of problems myself. It SEEMED like it had done some of the work for me.
This is a great AI prompt, Craig! Speaking as one of those “early career year 5 teachers” I would not have been able to generate such a variety of questions quickly, especially the proof examples.
I don’t think using AI will poorly impact the resource creation professional development of young teachers because they are still testing it the same way—on real students. Just like when I make a bad resource, give it to kids, and realize something is missing or confusing, the same thing will happen when teachers use AI generated resources. As the teacher, you still have to interrogate why students responded that way to the resource and how to improve it. The teacher needs to do that thinking for themselves before they can go back to AI or adjust the resource.
"But is that messy, inefficient, poor-quality resource creation a necessary process teachers must go through to build up their knowledge and experience of what an effective resource is?" - This is the very crux of the entire AI question. The short answer is yes. I recall a statement from Vara in her talk at Stanford's HAI earlier this year: "The goal for a writer, after all, isn’t for their perspective to magically appear on the page through no efforts of their own; the point *is* the effort of figuring out what it is you’re interested in articulating.” Across all domains and disciplines, something "learned" is the ability to fluently apply/deploy/perform, which is impossible to achieve without earlier miss-steps, dead-ends, reworking, walking away, throwing everything in the bin and starting over, being exposed to alternative approaches, reflection, synthesis with intuitions and other local factors, and then still not being happy with one's application/deployment/performance.
No, I think talking to AI this way forces teachers to think much harder about the learning process and student misconceptions. This will accelerate their progress as teachers. I love the way you keep returning to test AI capabilities.
There are two things that are on my mind after reading this.
1) we have to test AI created resources long term to see whether it works - my instinct is telling me that there will be certain repetitiveness of the design.
2) as a teacher I have to read and consider many incorrect solutions. Using AI expects me to fact check again. So I will be reading a material that is incorrect. It may be actually easier to create a resource. Not saving the time. It is more about the mind being exhausted from constant task switching and focusing on creating a task (without AI) may allow my mind to focus in a different (undisrupted) way.
Hi I read this article with interest. I am not a Maths teacher in fact I teach Religious Studies Ethics and Philosophy. I liked the Smith’s rubric in the article and asked Gemini to use the article and the rubric to create a worksheet that I could use as an end of topic task for one of the topics in GCSE RS Ethics. I was pretty impressed with the results as a first attempt.
I have been playing around with this as a primary school maths teacher/leader and I have to say I think it is fantastically simple. The opportunity to demonstrate within the model as well create something more child friendly is also incredibly easy to add. I love how simple it is to change and vary the outcome to better match what I need. For example, I made a rounding to 10, 100 and 1000 sheet. I asked for a snazzier title, more supports (like numberlines), more questions for question 2 and larger font. Done. In seconds. But, the beauty is each new iteration can use the references and refinements I have already made. Very much something I will be looking to explore more in the future.
Question 3a in the Triangle area example has some interesting four-sided ‘triangles’‽
What I want to know is why can't the pro versions of AI solve the "produce nice-looking mathematical expressions and images natively" problem yet? And don't get me started on diagrams. Not even ChatGPT5 can get those right. I've found that by the time I correct all the errors that AI-produced maths questions are riddled with, it takes about the same amount of time as creating them from scratch myself.
I asked Chat GPT to make me a graphing inequalities worksheet the other day, with graphing axes, and it gave me what appeared to be text based axes, rather than images. I assume it's just a limitation of the platform of which I was not aware since I am new to the use of AI. I ended up just making a worksheet but it was nice to have the list of problems generated by AI, even though I could probably just have thought of problems myself. It SEEMED like it had done some of the work for me.
This is a great AI prompt, Craig! Speaking as one of those “early career year 5 teachers” I would not have been able to generate such a variety of questions quickly, especially the proof examples.
I don’t think using AI will poorly impact the resource creation professional development of young teachers because they are still testing it the same way—on real students. Just like when I make a bad resource, give it to kids, and realize something is missing or confusing, the same thing will happen when teachers use AI generated resources. As the teacher, you still have to interrogate why students responded that way to the resource and how to improve it. The teacher needs to do that thinking for themselves before they can go back to AI or adjust the resource.
"But is that messy, inefficient, poor-quality resource creation a necessary process teachers must go through to build up their knowledge and experience of what an effective resource is?" - This is the very crux of the entire AI question. The short answer is yes. I recall a statement from Vara in her talk at Stanford's HAI earlier this year: "The goal for a writer, after all, isn’t for their perspective to magically appear on the page through no efforts of their own; the point *is* the effort of figuring out what it is you’re interested in articulating.” Across all domains and disciplines, something "learned" is the ability to fluently apply/deploy/perform, which is impossible to achieve without earlier miss-steps, dead-ends, reworking, walking away, throwing everything in the bin and starting over, being exposed to alternative approaches, reflection, synthesis with intuitions and other local factors, and then still not being happy with one's application/deployment/performance.
Very curious as to whether or not AI can create satisfying open/investigative material. I feel it might (at the moment) find this a bit harder.