Thanks to Will Bailey-Watson, University of Reading history PGCE and member of HA Secondary Committee, for sharing his initial explorations of AI to support history teaching. Hopefully this will help to start a discussion about AI and history teaching. We’d love to hear from you!

Artificial Intelligence (AI) is clearly going to change the world. It probably already has done in ways I don’t know. So far, most conversations I have been involved in at an institutional level have focused on the negative implications for education – plagiarism, factual errors, lack of critical engagement and so on. Yet, from playing around with ChatGPT (a generative AI system, and other AI systems are available) and collaborating with a range of different trainees in different schools, I have already seen some clear benefits of early career and training history teachers using AI. These include:

• Generating ideas and starting points that help early career and training history teachers visualise what might be possible.
• Producing resources very quickly that help early career and training history teachers fulfil ambitious curriculum aims.

There is a caveat here: I haven’t seen or engaged with any debates about AI in history teaching. I might be behind the curve. However, in order to generate a discussion and share my experiences this term, I am going to focus on a few examples of how purposeful and critical use of AI can help early career and training history teachers produce resources that fulfil ambitious curriculum aims…while dramatically reducing workload.

Causation and consequence cards: Battle of Hastings and The Glorious Revolution

We are approaching the end of the Autumn term, when many of my trainees seem to be teaching about the causes of William’s victory in 1066. Once a trainee has decided that they want to get pupils to develop their causal analysis, one use of AI is to quickly generate ‘causation cards’ that help explain why an event occurred. However, when it comes to why William won the Battle of Hastings, many existing resources seem to focus on very short term factors, and neglect the wider European context. When supporting a trainee in their lesson planning, we asked AI if they could help:

Can you make ten cards on different reasons William won the battle of Hastings? To include: European context, Normans in Sicily and the papal banner.

By specifying some of the historical context that would allow for richer analysis, the resource became more bespoke for what the trainee wanted to achieve. Immediately it is clear how this resource could be used for a whole lesson. A class could read through the ten cards as a group, with the teacher teasing out how each card related to prior knowledge from the topic. They could then highlight direct and indirect causes that appear on the cards. They could group them together into factors that help explain William’s victory. They could identify links as part of a causal web and, of course, they could try to create a hierarchy of causes in terms of their relative importance to William’s victory. An experienced teacher might question why AI is needed to generate such a resource, but this sort of resource often takes ages for an early career or training teacher with developing subject knowledge to produce.

As a postscript, this works equally well for evaluating consequences, where the same process of making a resource to develop historical understanding can be achieved very quickly. This time to generate a starting point for a teacher teaching about the Glorious Revolution:

Can you make 6 cards showing 6 different consequences of the Glorious Revolution.

Here, the teacher could use the cards to compare immediate and longer-term consequences, or they could evaluate who would have benefited or lost out. Again, in seconds a useful resource is ready.

Similarity and difference of priorities and interests in Tudor England

As a teacher, I had a bit of a love-hate relationship with character cards. Some of my favourite, and most successful, lessons involved character cards. They would generate important discussions about generalisations in history, and the impact would be seen in subsequent lessons when the pupils remembered that not everyone was the same. However, they were such a pain to produce. Hopefully, AI can help us here. Recently, one of my trainees (Isabel) wanted to adapt her introductory lesson on How far was Elizabeth’s reign a ‘golden age’? She had already taught the lesson once, but didn’t feel it had sown the seeds required to make the most of the subsequent four lessons (where a different aspect of Elizabeth’s reign would be analysed to see if it supported the interpretation of a ‘golden age’). Isabel had established learning objectives: 1) identifying what different Tudor people would want from a ‘golden age’, and 2) comparing the priorities of different Tudor people if Elizabeth’s rule was to be seen as a ‘golden age’. It became clear that an appropriate activity would be to introduce a range of Tudor people and to consider what might be important to them in their lives. Enter AI:

Can you create six character cards about people in Elizabethan England. 100 words about where they live, what they do, their religious beliefs. They need to be 3 men and 3 women. At least two need to be poor. One needs to be involved in the arts. Reading age of 13.

Within seconds we had six characters who could be used to generate this discussion with the pupils. This resource would not be acceptable to some history teachers as the people are fictional, amalgamations of different people available to AI, who represent the issues and lifestyles that were common in Tudor England. There are questions here about historical authenticity. But, for a teacher who is satisfied that they can make this clear to the pupils, this resource does open up a host of opportunities, both in the specific lesson and for answering the enquiry question over time. For instance, revisiting the cards in each lesson to see whether each character cared about the issue being studied (e.g. artistic trends, military success, privateering etc), or producing a living graph at the end to see how much they cared about the ‘golden’ aspects of the reign, or indeed a final discussion about why some historians have such a different interpretation of the reign to the people who lived through it.

The changing size and strength of the British Empire

When teaching an overview study, I like questions that ask when did something happen? When did Britain become a democracy? When did the US Civil War become ‘inevitable’? When was the USSR closest to being a Marxist state? Among other things, these questions need chronological understanding and a grasp of change over time. Yet, they can intimidate early career and training teachers because they require a breadth of knowledge, and apt resources are often hard to find. In a recent ITE history session, a group of trainees were trying to identify overview questions about the British Empire, and Archie came up with ‘When did the British Empire begin, peak and fall?’ I thought this was an impressive suggestion because it takes three contestable stages of development and gets pupils to grapple with the periodisation of Empire through their developing substantive knowledge. In order to implement this, the teacher will need a range of possible periods that the pupils could argue to be the beginning, peak or fall. Therefore, this remains a pipedream for most early career and training teachers – producing the resources to help answer your enquiry question would be too overwhelming. Again, enter AI:

Could you create a five cards to use in a lesson about “The British Empire at Its Peak”. On each card there must be a different year that could be argued to be the ‘peak’ of the British Empire in terms of size and power. Each card needs about 100 words with specific details about the power of the British Empire at that time.

Straight away we have a basis for discussion. Pupils can build a sense of chronology through these dates, they can identify changes and continuities, and they can start to analyse the strength of arguments for and against each showcasing an empire at its peak.

However, and this is a big however, AI answers the questions we ask it. With the Hastings causation cards we added the specific context that we wanted, and AI was able to provide what we were after. With the Tudor character cards, we specified the diversity of people and the criteria we wanted to see. In this case, with the dates of the British Empire, insufficient thought had been given to what constitutes ‘peak’. Whose ‘peak’? AI had given what it asked for – cards that narrowly focus on the ‘peak’ as a manifestation of British power. Therefore, I needed to improve the question I asked:

Could you now include 50 words on each card which challenges the idea of the British Empire being at its ‘peak’ because of the suffering, violence, exploitation experienced by the people living in the colonies. For each of these cards, could you add a specific example of the negative impact of the British on a country within the British empire.

All of a sudden this lesson is really interesting. For each possible year of ‘peak’, the pupils need to examine who would challenge the basis of it being a ‘peak’. Remember, this is an overview introduction to the British Empire, and we have just smashed down a door that leads to a whole range of questions that allow us to unpick assumed norms in the language used to describe Empire.

Story telling: big stories such as the build up to WW1

Each year I use the story of Alphonse the Camel to show trainees all the different sorts of causal reasoning we can do, once we have secured a chronological narrative in the pupils’ minds. Every year, I have some trainees remember this session as being about Alphonse (i.e. about using mini-stories to teach small causal studies in individual lessons), rather than seeing that the same principles of teaching causal reasoning apply to whole topics and sequences of lessons – they forget that these are just bigger stories. A good example is the typical four lessons spent at the beginning of Year 9 on the causes of WW1. Loads of my trainees spend their first few lessons teaching this. And yet, despite our focus on securing that narrative first, like Alphonse, the causes of WW1 are often taught in isolated streams (e.g. militarism, nationalism, arms race, etc) which often means that many pupils don’t really achieve a secure narrative of the build-up to WW1. They are therefore doing their causal reasoning without that rock solid chronological narrative. I had a hypothesis that this is because it is really hard to find age-appropriate, readable stories about the longer-term build up to big events. And boy they would take ages for an early career or training teacher to produce. This seemed to be the perfect challenge for AI:

Write a 500 word summary story of the build up to outbreak of war in 1914. Start in 1870. In each paragraph could here be at least one specific date. Reading age of 12 year old.

This is a great example of where the teacher would need to read through the resource very, very carefully to ensure it is age-appropriate, touches on the substantive knowledge they want it to, and that they know exactly what want to do with it. However, suddenly, all sorts of opportunities are available. Perhaps the most obvious is to read through this in the first lesson of the sequence. This would easily take a whole lesson – unpicking the major themes, sequencing key events in a timeline, and laying a basis for subsequent explorations of the key causes.

Story telling: individual stories such as Sophia Duleep Singh’s life in London

One of my trainees, Emilia, and her mentor, Asha, are co-planning a short sequence of lessons on Sophia Duleep Singh’s involvement in the suffrage movement, as part of an enquiry: ‘To what extent did Britain become a more democratic society after 1745?’ Emilia decided that the first lesson would focus on world-building, with a lesson focus of ‘What does Sophia’s London reveal about the women’s suffrage movement?’ Emilia identified a range of places in London relating to Sophia and produced a lovely map for the pupils to annotate, on route to analysing what Sophia’s London revealed. However, the elephant in the room when conceiving this lesson is where will the content come from? When they want to plan their own lesson, trainees often end up cobbling together some online clips and burning the midnight oil by creating their own information sheet. They could continue to do this but, you know, time… Exercise, TV, friends and sleep are important too if we want people to continue to enjoy teaching. So could AI at least provide a starting point:

Write a story about Sophia Duleep Singh. 500 words. Must focus on her actions in the suffrage movement and refer to places that were important to her in London. Reading age of 12.

Fortunately, Emilia noticed a glaring error – Singh was not incarcerated in Holloway Prison, and AI has presumably confused Singh with another suffrage campaigner such as Rosa May Billinghurst. Or it has simply hallucinated. As the course leader, I was delighted, as this is the sort of criticality that I want my trainees to show whenever they consider utilising YouTube clips, online worksheets, or anonymous resources found in old department folders. However, once the story had been critiqued, it was evident that, within 15 seconds, we have the starting point for a resource that potentially supports a really innovative lesson. Again, it is important to read these stories carefully, to ensure the content is accurate, to check that it hits the curriculum buttons you need it to…but this is true for any resource a new teacher finds on their school system or the wider internet. With these stories, it is amazing that you can adapt the length, reading age, detail included and, most fun of all, request for the stories to be written in different styles (e.g. using evocative language, decolonising this person’s story, etc).

Once in the generative AI headspace, it is hard to stop seeing ways AI can help. For instance, I have always hated producing multiple choice questions – good ones take ages – so it is great to see how AI can help so long as the terms are made really clear. Or, how about, after studying 1919-1939 in Year 10, you can ask AI to produce an alternative 1919 peace treaty and start the lesson by asking: would AI have done a better job that the Big Three? Or after studying WW1 (or literally any topic), you can ask AI what was the most significant event in WW2, and conclude the sequence of lessons by asking: do you agree with AI about what the most significant moment of this topic was? (In case you are bored of clicking links, AI says that D-Day was.) Or, finally for now, after studying 1917, you can ask AI to produce counterfactual questions that could have changed the course of the Russian Revolution, and begin a lesson by discussing these questions.

You will note that this is not an article with ground-breaking resources. Hopefully, there will be a blogpost or article very soon that shows the true potential of AI. When I told my friend Charlie that I was writing this blogpost, she told me of her idea that teachers could set a task whereby pupils use AI to answer a history question, and would be assessed on the questions they ask AI. Her hypothesis is that many people in education are scared of AI because they presume that pupils have the critical literacy and historical enquiry tools to ask it the questions that experts ask. Needless to say, I love this idea. This example shows a transcript of questions when the task was: To what extent does AI make convincing and supported arguments about the causes of WW1? You can see that the dialogue with AI shows that the user is reading the responses carefully and trying to ask questions that a historian might ask. Surely someone wants to give this a go?

Furthermore, it will be game-changing when AI can help assess pupils quickly, generate detailed feedback and generally make teachers’ lives easier while potentially improving the quality of our practices.

As I established at the beginning, this blogpost has looked at how AI can help generate ideas and starting points, and help produce resources very quickly once the teacher knows what they want to achieve. Nothing about my experiences with AI suggests that it will replace the need for highly knowledgeable, well-trained teachers, who know what to ask and how to implement pedagogies. Central to my excitement about the potential for AI is that, so far, I have seen AI can help early career and training history teachers to be more ambitious, and achieve greater autonomy, as it allows them to generate ideas and produce resources much quicker. I have only seen AI be used effectively in the classroom when the teacher knows their curriculum aims and thinks carefully about how the ideas and resources will be made to work with their pupils. As a result, developing the critical literacy of my trainees has already become an important part of my ITE curriculum. Therefore, despite me starting small with simple suggestions, for stretched early career and training history teachers who want to be more ambitious, who want to achieve different sorts of aims, and who don’t have endless amounts of time, I really do think there are opportunities for experimenting with and harnessing the potential of AI.

To find out more you could listen to this EuroClio podcast from the HU University of Applied Sciences in Utrecht. Two academics discuss AI and history teaching.