Leon Furze

Across Australia and internationally, Departments of Education, School Associations, and individual schools and teachers are asking the same question of generative AI: can it solve the teacher workload problem?

Over the last 18 months, I’ve seen literally hundreds of ideas discussing how AI can assist with everything from administrative tasks to classroom activities, communications with parents and community, and everything in between. Unfortunately, much of this advice has come from outside of education itself – tech developers and companies on the fringe of education whose ideas would seem attractive, but which for reasons I’ll discuss in this article might not always be a good idea.

Educators who are using generative AI technologies are producing fantastic materials and, in the spirit of teachers everywhere, sharing them widely on social media and through professional associations. But those teachers are still few and far between. Every school might have a handful of educators who have the time and the inclination to dive into generative AI, but the majority of K-12 teachers and university educators that I’ve worked with simply haven’t explored the generative AI enough to make big gains in handling their workload.

Understanding the problem

In the last few years in Australia, there’s been a lot of media attention and political focus on teacher workload, largely due to issues with teacher retention, burnout, and low numbers of students studying teaching at university.

It’s easy to see why Generative Artificial Intelligence has been positioned as a solution to some of these bigger picture problems, with the assumption that if we can dig teachers out of the avalanche of administrative tasks and other duties outside of the classroom, and free up more time to actually teach, then people will be much more likely to enter the profession and stay in it for longer. There’s plenty of research out there about what constitutes teacher workload and why many teachers feel so overwhelmed.

This recent synthesis of research by Creagh et al., for example, explores the “time poverty” of educators, including teachers and principals. It offers a few key observations about teacher workload, including:

1. Studies commonly report teachers working over 50 hours per week, with work spilling into weekends and personal time.
2. While some research suggests teachers’ working hours have remained constant over 25 years, most studies report workload has increased due to the intrusion of “non-core” tasks like administration, curricular reform, and compliance with accountability frameworks.
3. Principals also report high workloads of 50-60+ hours per week encompassing all facets of school management, legal issues, marketing, pursuing alternative funding, and extensive stakeholder engagement. Non-educative work is seen as impacting principals’ capacity for educational leadership.
4. Technology has exacerbated principals working outside traditional hours because of a perceived need to be more available and visible. Administrative work gets pushed to evenings and weekends as a result.

Unfortunately, research into teacher workload is sometimes misconstrued, or used to serve particular agendas. For example, the Grattan Institute in Australia has produced reports indicating that planning and resource creation are one of the primary burdens on teachers, and their solution is to provide Departments of Education teachers with pre-prepared unit plans, lessons, and resources.

The announcement was met with a lot of pushback from teachers, who stated they would rather have time than be handed resources. Aligned to this attitude towards workload, however, it’s easy to see why so many providers are developing applications designed to produce lesson plans and resources for teachers en masse. The approach appears intended to address issues like educators who are teaching out of field (not in their trained subject area), newly graduated teachers who find themselves entering a school that is not well resourced, and experienced teachers who feel they do not have the time to develop quality resources.

Based on my own experience as a teacher and council member for the Victorian Association for the Teaching of English, and regular conversations with teachers, most don’t want off-the-shelf resources to take into the classroom. Like the comments made in the Conversation article above, they want the time, space, and collaboration to create rich and meaningful resources that lean into their areas of expertise and interest. In my experience, they certainly don’t want cookie-cutter lesson plans from generative AI or anywhere else.

So, if providing educators with off-the-shelf resources isn’t the answer, how can generative AI used effectively reduce teacher workload?

Practical AI Strategies is available now from Amba Press

Understanding how the time is spent

In March 2024, I worked with Independent Schools Victoria principals to understand the implications of generative AI for their roles and for their schools. On the day, one of the activities we tried out was the classic Eisenhower matrix, sometimes called the urgent-important matrix. This approach to tracking time was popularised by Stephen R. Covey in his book The 7 Habits of Highly Effective People.

The approach divides a grid into four quadrants, with tasks ranging from not important to important and not urgent to urgent. These quadrants then provide four ways of dividing up day-to-day activities:

• Quadrant 1 (Urgent & Important): These are tasks which are of high value and need to happen right away. For example, emergencies which must be dealt with by the individual, priorities which absolutely cannot be delegated or pushed aside, or high-value interactions which have a near deadline.
• Quadrant 2 (Important & Not Urgent): These are tasks of very high value which do not have the pressing urgency of quadrant one activities. For example, long-term plans and strategies, activities which will only be rewarded in the future, and tasks which are beneficial to sustaining an organization or an initiative but which require significant planning and scheduling of time.
• Quadrant 3 (Urgent & Not Important): These tasks are often other people’s problems which have either been delegated to the individual or which have descended from an external organisation, such as compliance or accreditation. Deadlines have to be met, but the activity is not necessarily important for the individual’s day-to-day work.
• Quadrant 4 (Not Urgent & Not Important): These are time fillers or time wasters – activities which don’t add anything meaningful for the individual’s role or the people that they work with.

It’s important to note that there’s nothing inherently wrong with any of the quadrants, although Covey would recommend that we try to shift most of our daily tasks into Quadrant 2, meaning we are spending time on longer-term, meaningful, and impactful work and less time putting out fires, responding to other people’s business, or wasting time.

Sometimes it might be absolutely necessary to kill half an hour during the day scrolling through social media, but if that’s a daily habit, it’s probably not very helpful in the long run. In order to get most activities into Quadrant 2, it’s necessary to be proactive – to plan ahead for problems before they crop up, to identify appropriate ways to delegate or reassign responsibilities, or to automate or make more efficient some of the more pressing but less important work. Generative AI can potentially help at all of these levels.

The following example is just based on my experience as an educator. I’ve populated the four quadrants in ways that I believe are appropriate, but you might disagree with where I’ve placed some of these things. That’s fine. Ideally, if you like this idea, you should carve out half an hour to sit down and do it yourself. Reflect on all the things you do day-to-day or week-to-week. Write them down in a long, unfiltered list, and then divide them up into the four quadrants as you see fit.

Urgent & Important	Not Urgent & Important
– In-the-moment feedback – Important scheduled meetings (e.g., with parents) – Behaviour incidents – Today’s lessons – Assessments and reporting deadlines – Physical injury or mental health crisis – Responding to urgent communications from school leadership – Responding to genuinely important communication from parents and students	– Personal physical & mental health – Own career goals – Longer-term subject & unit planning – Faculty-level strategic planning – School-level strategic planning – Professional learning – Developing skills outside of professional context – Building positive relationships with students, colleagues, and community – Scheduling time for important collaboration
Urgent & Not Important	Not Urgent & Not Important
– Phone calls and most emails – Surprise visits (e.g., from students at lunchtime) – Most administrative and compliance deadlines – Covering classes for absent colleagues – Dealing with technology issues (e.g., broken wifi router) – Attending unproductive meetings – Completing unnecessary paperwork to a deadline	– Social media use – Unproductive conversations – Time-wasting activities which don’t contribute to any of the other quadrants – Compulsive email checking – Procrasticleaning – Over planning lessons or resources (or laminating anything, ever.)

As I said, there’s nothing inherently wrong with any of the tasks in any of the quadrants, but you can see how if you spend most of your time in Quadrant 3, dealing with other people’s problems or responding reactively to situations as they crop up, then you’ll have far less time for the important work in Quadrant 2. There’s even a time and a place for Quadrant 4 activities. Sometimes you just need to get something off your chest, wind down for an hour, or vent to a colleague. But generally, unless you turn those vented frustrations into something more proactive, they don’t lead to anything beyond a bad mood.

Shifting the dial

Before we’re even talking about generative AI, it’s worth looking at how some of the issues can be moved towards Quadrant 2. For example, if you find you spend a lot of time in the classroom managing behavioural issues, there is obviously a need for both a personal and a school-level strategic approach to behaviour. I absolutely do not hold with the idea, which seems to be prevalent in the media at the moment, that teachers are solely responsible for shaping the behaviour of young people.

I also reject the implications of suggestions for things like a “behaviour curriculum,” which abrogates the responsibility of society as a whole for helping young people to grow and develop in ways which aren’t harmful or hurtful. Frankly, the idea of a set of worksheets that some poor homeroom teacher has to go through with a group of 20-something adolescents to teach them how to behave properly fills me with horror.

But there are, of course, ways that you can devote quality time to improving your own skills and thinking strategically that will, in the long term, pay off. For example, spending half an hour during a free period or over lunch reading through a well-researched and inclusive book like Christopher Hudson’s Leading Positive Classrooms is a decidedly Quadrant 2 activity that might ultimately reduce the time spent managing behaviour in Quadrant 1 (Chris and I share a publisher, but even if we didn’t I’d wholeheartedly recommend his book).

Similarly, if you find yourself spending a lot of time during the week resourcing lessons on the fly, that’s worth identifying ways that you, your faculty, and your school can support more conscientious planning.

Particularly for those of you in middle management and executive management positions, I imagine you spend a lot of time in Quadrant 3, dealing with other people’s problems. Empowering staff and the community at every level to solve their own problems is a long-term Quadrant 2 activity – through professional development, work with parents and friends, and so on – but the more time you invest in encouraging people to solve their own issues, the fewer of those issues are going to land on your desk unexpectedly in the middle of the day.

None of what I’m saying here should be news to anyone, but it is something that requires a great deal of conscious and collaborative effort across all levels of the school.

Identifying where AI can help

Using the Eisenhower Matrix as a framework, I’m going to explore a few ways that generative artificial intelligence can be used to support teacher and school leader workloads beyond generic lesson planning and resource creation.

Quadrant 1 Activities: GenAI On-the-Fly

GenAI is quickly becoming ubiquitous, and this could be a good thing. Having Copilot, GPT, Claude, or whatever, always on hand can be a boon when something comes up that is urgent and important.

For example, often in the middle of a lesson, particularly with senior students, conversations can become diverted down rabbit holes which are interesting (for the students at least) but perhaps outside of the teacher’s areas of expertise or not appropriate for the content of the lesson. These are often some of the most engaging (for the students) moments in the class – they’re not planned for, they’re spontaneous, and they’re often predicated on the kinds of relationships with students that teachers value. Rather than brushing off these urgent and sometimes important conversations in the classroom, artificial intelligence can be used both to explore these ideas and to bring the lesson back on track.

Here is a concrete example. You’re in an English classroom studying a set text, say The Crucible, for a major assessment or final examination. In the middle of a discussion about the relationship between two of the characters in the text, a couple of students make a comparison to a popular reality TV show (if you’re in Australia, you’ll know what I mean when I say MAFS), and the quality of conversation in the classroom quickly dissolves.

Rather than trying to stanch the flow of conversation or spending too much energy dragging the conversation back on track, use an Internet-connected AI to draw a comparison between what the students are talking about and the text that you’re studying. Write the results up on the board, and then move on.

How this helps with teacher workload

This might seem like a facile example, but a lot of time is wasted in the classroom battling to get students on topic when they’re determined to talk about something else. This can be stressful for the teacher, can lead to further behavioural issues if the relationship becomes antagonistic, and can create more work if content isn’t covered in one lesson and then has to be revisited in the next. And just in general, it can be stressful and annoying. But using generative AI in this light-hearted approach shows the students you care about what they’re interested in, but also that you’re only willing to discuss that within the bounds of the lesson. You can keep things on track and allow yourself to move on without sacrificing those relationships.

Even better, you might incorporate some of the generative AI results into your planning for the next time you teach this topic. Although, by the time you’ve got another group of kids in front of you in 12 months, it’s highly likely they’ll be interested in something totally different and equally distracting.

Hand on the door-handle planning

Another Quadrant 1 activity example is the “hand on the door-handle” lesson plan. At some point in anyone’s teaching career, they do what my mentor all the way back in the early 2000s used to call the “door-handle lesson plan.” Maybe you’ve been under the pump all week – it’s one of those weeks where you have to also cram in swimming sports or athletics, parent-teacher interviews, and so on. Maybe it’s in a heavy assessment or report writing season. Or maybe (don’t pretend this has never happened to you) you actually just lost track of which day it is and had no idea you’re teaching this class.

As you approach the classroom, you place your hand on the door handle and hope that inspiration strikes. This is definitely Quadrant 1 material – it’s urgent (you’ve literally got your hand on the door handle) and it’s important because if you don’t think of something quickly, you’re going to have a bunch of fairly bored adolescents staring at you.

Here is an opportunity where generative AI can be useful. We still don’t need generic lesson plans, but if we can quickly write a prompt which has detailed enough context, we can get something that meets the needs of this particular lesson very quickly. Here is an example:

For some reason, it slipped your mind that in less than 15 minutes, you have to teach a Year 10 Science class on biofuels. You flip open your laptop and use the following prompt:

I am teaching a lesson on biofuels to a Year 10 Science class. Please provide the following:

1. A brief introduction to biofuels, including what they are and why they are important (2-3 sentences).
2. Three key points to cover about biofuels, with a one-sentence explanation for each point.
3. An engaging, hands-on activity or demonstration that can be completed using common classroom materials to illustrate a concept related to biofuels.
4. Two or three questions to ask the students to check for understanding and encourage discussion.
5. A one-sentence conclusion to summarise the main takeaway about biofuels.

Use the following information to provide context: <copy/paste a curriculum document or a resource from the school Learning Management System>

Don’t make a habit of it, but in a pinch, GenAI can produce contextualised, specific lesson plans in a hurry.

Not that this ever happened to me, but I probably could have used this a few times in my career to save the day and reduce my Q1 ‘urgent-important’ workload. Obviously, I don’t recommend you teach every lesson in this fashion, but you can see how this will help in a pinch, reducing stress so that the lesson is still meaningful and on-topic. It still uses your existing resources or input from the classroom, and it solves the immediate problem. As with the earlier activity, you may even end up creating a really interesting and engaging resource that can go into your longer-term planning.

Quadrant 3: Other People’s Problems

A lot of the time, Quadrant 3 things revolve around electronic communications. An email suddenly lands in your inbox from a parent. It’s urgent and important to them, and if your school has a 24-hour turnaround policy, it’s urgent enough for you. But you can’t let it become a time sink. Generative artificial intelligence can be used to understand emails, particularly ones which are longer, complex, or fraught with emotions, and then produce draft responses to those emails which don’t require a great deal of mental energy.

Increasingly, this is being built into popular email platforms. For example, you can use Google’s AI in Gmail and Microsoft’s Copilot in Outlook, if you have access to Copilot for Microsoft 365. But this shouldn’t mean we end up in an endless cycle of AI talking to AI, where you use your chatbot to write an email, somebody uses theirs to respond, and so on. As with all of the Quadrants 1, 3, and 4, we should be looking for ways to shift these tasks towards Quadrant 2 tasks that reduce the overall volume of these communications.

But in a pinch, using AI can reduce the time taken to respond to this type of communication. For external deadlines, responding to other people’s deadlines, requests for information, and so on is unfortunately part of daily life. It’s urgent, carrying a specific deadline, but in the grand scheme of things, it’s probably not important for the individual. There are three ways to deal with these with generative AI:

1. Use generative AI to help understand the requirements and produce the content.
2. Use generative AI to draft the content to send back.
3. Speak to whoever is in charge of issuing these things and ask them to please stop.

Quadrant 4: Time Wasters

If you find yourself endlessly doom scrolling social media or standing in the staff room talking to anyone who’ll listen about how frustrated you are every single day, then, respectfully, you’ve probably got bigger problems than AI. In my experience, people don’t actively avoid their job unless they’re particularly stressed out, antagonised, threatened, diminished, or otherwise unhappy or burnt out.

Quadrant 4 is not a good place to spend most of your time. It’s obviously unproductive – throwing hours of your workday into TikTok isn’t going to hit any goals no matter how many #TeacherTok videos you scroll past. But even worse than that, you’ll finish off the day feeling unproductive, with that complex mix of frustration and shame that comes from knowing you haven’t done a good job and not really knowing how to get yourself out of that position.

In this instance, generative artificial intelligence isn’t the answer. If it’s a one-off, fine. If it’s something that’s happening daily, talk to someone you’ve got a strong relationship with, such as someone in school management, and tell them how you’re feeling. If you don’t have someone you can trust in your organisation, find out if your school or sector has counselling services for teachers. Failing that, speak to someone in a public mental health organisation like Beyond Blue in Australia.

Obviously, generative AI is not a panacea. If you’re in a temporary rut, though, then you could use generative artificial intelligence to get yourself out of the ditch. For example:

I’m finding myself wasting way too much time on social media at work. Suggest three evidence-based strategies for getting me back on track and being more proactive.

Quadrant 2 activities: Longer term workload reduction

This is the most important quadrant, not only because it’s where we’re getting the biggest workload gains from using generative AI effectively, but also because despite the idea of generative AI for solving low-hanging fruit, quick lesson planning or resource creation is not where the real power lies.

Quadrant 2 activities centre on projects which require us to create space, dedicate time, and create opportunities to step back from the urgent demands of the job and deliberately focus on our roles and our future goals. Here are some suggestions on how generative artificial intelligence can help:

Automate the Boring Stuff

This is the title of a series of books on coding by Al Sweigart, and automation is a key facet of his programming philosophy that should be part of your day-to-day work. If you find yourself doing something again and again and again, chances are there is a way you can use generative AI to automate it. This can range from simple tasks, like reformatting documents (e.g., “Take these notes and expand them into Understanding by Design lesson plans with the following sections…”), to much more complex ideas, depending on your level of expertise with the technology.

For example, if you find yourself constantly reformatting documents, taking notes and putting them into pro-formas or templates, you can see if there’s a way you can use generative AI to take uploaded notes and export them as an Excel file. These more advanced tactics will mostly only work in more sophisticated, paid versions of models like GPT-4, but it can be worth the 20 US dollars a month. I’ll be writing much more in the future about automations for educators.

Strategic Planning

I’m a huge strategy nerd, with a bookshelf full of saccharine-sweet business and self-help books from which I cherry-pick the ideas that I like the most. Most sensible people probably dedicate very little time to strategic planning, unless you’re in a leadership role in a school. Here, you can easily use generative artificial intelligence to skip some of the learning process and help step you through quick and effective activities that can help you to set goals, plan units of work, or build bigger picture faculty and school strategic plans.

Investing a little bit of time in this kind of thing short-term is going to pay huge dividends in the long term, helping to solve problems like the “hand on the door handle” planning, or perhaps even mitigating some of the risks of behaviour incidents by identifying potential problems. For example:

Here’s a unit of work. Identify places where students might become disengaged, where roadblocks could occur, or where the volume of work might cause problems for students completing tasks. <copy/paste or upload unit of work>

Or this example:

I graduated as a teacher two years ago and I’ve been working full-time as an English teacher. I’d like to progress my career, expanding first as a faculty coordinator and then I think into senior school leadership such as a deputy principal. Ask me some questions that help me develop a plan for that professional learning and development over the next 12-24 months.

Professional Learning Networks

Building a strong professional learning network (PLN) is can be important for long-term growth and development, and also puts you in touch with some great like-minded educators who can help you if you’re starting to feel bored, listless, or burned out. Internet-connected GenAI (like Copilot, Gemini, or ChatGPT Plus) can help you identify key people, resources, and communities to connect with based on your specific interests and goals.

For example, you could use a prompt like:

I’m a secondary school science teacher interested in project-based learning and integrating technology into my classroom. Suggest five key educators, bloggers, or thought leaders I should follow on Twitter or other social media platforms to learn more about these topics.

Or:

I’m a primary school principal looking to build my knowledge of inclusive education practices for students with diverse learning needs. Recommend three professional associations, online communities, or conferences where I can connect with other school leaders and experts in this field.

Data Analysis and Informed Decision Making

It’s easy to get overwhelmed by the sheer volume of student data available from standardised tests, to student sentiment monitoring platforms, to self-assessment and reporting. GenAI can help teachers and school leaders make sense of this data, identify patterns and trends, and use insights to inform instructional decisions and school-wide initiatives. These can all contribute to more proactive Quadrant 2 approaches to planning units of work or building relationships with students.

For example:

I have a spreadsheet of my students’ scores on the last three formative assessments in my <subject> class. Identify any patterns or trends in the data and suggest two or three potential interventions or instructional strategies I could use to support students who are struggling with specific concepts.

Always de-identify data before uploading to GenAI!

Or:

Our school recently conducted a climate survey of students, staff, and parents. Based on the data in the attached summary report, what are the top three areas of strength and the top three areas for improvement? Suggest one specific initiative or strategy we could implement to address each area for improvement.

Conclusion

Using generative AI to reduce teacher workload requires thinking beyond the low-hanging fruit. We don’t need generative artificial intelligence to write lesson plans or create off-the-shelf for us. That’s what we’re trained for, that’s what we’re experts in, and in my experience, that’s what teachers want to do. We need ways to maximise the time that we have for meaningful, collaborative meetings and working with one another as a community.

Over the next few months, I’ll be exploring more of these teacher workload ideas, and speaking with educators in K-12 and tertiary about sharing how they use GenAI to reduce their (not)urgent and (not)important tasks.

If you’ve got examples you’d like to share which have genuinely had a positive impact on your workload, please get in touch using the contact form below.