Talking to Teachers About AI in Schools
For the past couple of weeks I’ve been travelling between New South Wales, Victoria, and South Australia and working with schools for their start-of-term professional learning. It’s probably the part of the job I enjoy the most, as it puts me in touch with a diverse group of educators who are all struggling with the same problems: how to understand AI in schools when it seems to move so quickly, and how to make sure that the students, above all else, are learning.
This post is a reflection on those school visits, with some of my own assumptions from fifteen years in the classroom thrown in. There are no clear answers yet about whether AI harms or helps learning, or whether its a net positive or negative for society as a whole. There are positions on both sides of the fence and just as many educators resisting and refusing as there are experimenting and using AI. One thing is consistent though: educators are keen to understand the implications of AI for themselves and their students.
The Big and Small Questions
Over two years on from the release of ChatGPT, the big existential questions about AI are still the most common. Will AI harm critical and creative thinking? Will it improve equity of access to knowledge and content? What skills will students lose? What will they gain? Will the benefits of the technology outweigh the costs, such as the environmental concerns? How do we deal with the increased corporatisation of education, and technology companies’ influence in schools?
The very fact that these questions are still the most asked shows that teachers are keeping an eye on the bigger picture, and absolutely not shying away from the problems and the possibilities of AI in schools. But teaching is also a frantic, intense job. These big questions will almost certainly give way over term one to other priorities, including curricular and extra-curricular activities.
Alongside these questions are some which are perhaps more mundane, but just as important. What is Copilot and why is it suddenly reading my emails? Did I see somewhere that ChatGPT can produce PowerPoints? How does that work? Which AI is the most reliable for research? If a student uses it for research, is that different from Google? Can I even set homework any more?
Teachers are looking for answers to the big and small questions, and unfortunately they don’t seem to be finding them. It’s impossible for anyone – even full time researchers – to keep up with the developments of AI. Between OpenAI’s weekly feature drop, Google’s slipshod approach to releasing apps, and Microsoft’s infuriating Copilot licensing structure, it’s unlikely that any individual teacher, let alone a whole school, can fully plan for this technology and its implications in the classroom.
Where Do Schools Turn for Advice?
In almost every school I work with, one or two people are shouldering the burden of providing information to staff on “the AI thing”. At one school, it was an IT administrator. In another, the Deputy Principal of Teaching and Learning. In a third, the Digital Technologies teacher. Often, anyone with the word “digital” in their role becomes the de-facto AI person, fielding all of the questions above, big and small.
Outside of schools, various institutions are attempting to provide knowledge bases for staff. Independent Schools associations, Catholic diocese offices, and the Department of Education in various states are all working separately on resources, videos, explainers, and advice. But by the time these large organisations have approved a project, made the resources, and circulated them to the schools, teachers are finding that they’re too late.
The government, thus far, hasn’t been particularly helpful. Projects investigating AI in schools have focused on “reducing workload” for teachers, a productivity mindset which treats things like lesson planning as low-hanging fruit. But when I speak with teachers, they don’t want AI that writes their lesson plans. They also frequently don’t want – or trust – AI “tutors” or AI that churns out worksheets.
We need to make more effort to listen to teachers and find out what they do want. A large part of this is understanding that these educators are the experts in their subject discipline, and their classroom practice. I’ve written before about the “expertise problem” with AI: you have to be an expert in your discipline to get the most from the technology, but you also need a certain level of proficiency in the technology itself to get the most from it.
A skilled science, maths, or computing teacher working with a product like OpenAI’s new o3 reasoning can probably do great things. An experienced English teacher might find o3 useless, but have a dozen approaches to getting the most out of Claude or GPT-4o for helping students assess and edit their work. That level of understanding will never be possible if teachers don’t get enough time to experiment and learn.
I’m not an AI Teacher, I’m an English Teacher
The teachers I speak to are not uninformed, and they’re not anti-technology. When they have concerns about AI, they’re based on genuinely problematic aspects of the technology we should all be concerned about. And often, the biggest push back is that at the end of the day teachers are subject matter experts, not AI experts.
I think that if we’re going to tackle the gaps in understanding of what AI can and can’t do in education, we need to start from there. Not “what are the impacts of AI on learning?” – a broad and ambiguous question that makes the assumption that “learning” is a fixed thing that can be measured. But: what does this kind of AI mean for the Sciences? Or the Arts? Or the Humanities?
I show Media Studies teachers image, audio, and video generation, and they are immediately brimming with ideas about student projects, rapid prototyping of films and animations, analysis of image-based biases, and the discussion of corporate control of the media, which now includes AI. I show science and maths teachers the reasoning and “thinking” processes of models like o1 (and now o3), and they understand how it might be used to show a student how to think through a complex task step-by-step, and how to reverse engineer problems.
Every discipline will respond differently to different aspects of the technology.
The “Discoverability Problem” of AI in Schools
Compounding the expertise problem is something software designer Patrick Hebron calls the “discoverability” issue of AI. Basically, AI doesn’t come with an instruction manual, or even a handy toolbar. Microsoft Word has a “ribbon” of features, including style – bold, italics, headings – but also more advanced features like mail merges, review tracking, and so on. ChatGPT, on the other hand, looks like Google.
That’s a deliberate design choice to make it feel familiar and user-friendly, but it’s awful for figuring out what AI can actually do. It’s impossible to tell from this little text box that ChatGPT can, for example, write and execute its own code to produce downloadable files in common formats like documents, PDFs, spreadsheets and PowerPoints. You can’t tell from this Google-ish interface that GPT-4o can do a decent job of analysing data, or that any of these models can read and write scientific formulae.
To figure out the potential of a product like ChatGPT – to discover its features – is a work of trial and error, and the kind of persistence that most teachers simply don’t have the time or the inclination for.
My discipline is English, so I can think of dozens of ways an English teacher might use AI to assist, rather than replace, students’ writing. But I don’t know the full extent of how AI can help in other disciplines. We need subject matter experts across education to contribute.
If you’re involved in a Maths Association, or Science, or English, or any subject discipline, that would be a great place to start. Speak to your colleagues, and start to build up a level of expertise in using AI. Share those resources around at association conferences and online. Don’t wait for generic AI conferences – often tech sales pitches anyway – or government resources which will be inevitably broad.
We need subject matter experts to help discover what AI can do within their discipline, both from the perspective of helping teachers and the implications for students and learners. Outside of the classroom we need finance, business managers, executive assistants, registrars, and other administrative staff to put their heads together and work through the implications of AI in schools through their own domain expertise.
All of this will take time, but whether we like it or not this technology is not going anywhere. We have time, but we need to invest it in the right place. From what I’ve seen in the past few weeks, that means handing the reins to the experts, and letting them tell us how to use AI in schools.
Want to learn more about GenAI professional development and advisory services, or just have questions or comments? Get in touch:
Leave a Reply