The AI Advantage: Streamlining School Leadership with Generative AI
Artificial intelligence… It’s utterly terrifying or it’s jaw-droppingly impressive. What camp are you in?
Elon states that we are ‘summoning the demon’. Hawking sat on the fence: “It will be the best or worst thing ever to happen to humanity, we don’t know which yet”. Demis Hassabis, the prominent figure in the AI world said that “I believe AI will be the most important technology of the 21st century. And it will be for the benefit of all of us if we get it right.”
I’ve spent the last few months looking at how education leaders use AI across Europe and America. What really struck me was that most of the really innovative stuff is going on elsewhere. The bus has left without us. A few weeks ago, I met a group of German exchange sixth form students on an engineering visit. AI was underpinning most of their work. For example, they studied how AI solved an engineering problem, then applied the same principles to an unfamiliar problem which they solved collaboratively. Meanwhile their British counterparts were being lectured on how, if they dared use AI, they’d be accused of plagiarism or how the exam board didn’t allow it.
Teacher training in the UK limps behind too. AI is not mentioned in the teacher standards. Teacher training providers generally do not include AI within their training modules so trainees’ exposure to AI is largely dependent on whether their school mentor takes a personal interest in it. So, while we grapple with the ethical and practical challenges of AI, Germany is funding AI pilot projects designed to weave it into the fabric of education.
What is generative AI and how does it work?
If you’ve seen Terminator or War Games, you get the idea. The machine learns from the information you give it and gradually gets smarter. Machines rule. Your mother teleports from the past and saves us all from Armageddon.
Generative AI has recently become much faster in its learning due to something called Large Language Models (LLMs). LLMs are trained on books, poems, texts, news items and other media. This enables them to learn the patterns, structures and nuances of language. If you ask it a question, it will understand nuance and context, providing a coherent answer (well, most of the time!).
There are a few Generative AI tools on the internet which include LLMs. The most well-known is ChatGPT from OpenAI, which is the Big Daddy of AI. The free version is an ideal place to dip your toe into the world of AI. Then there are education specific sites. TeachMateAI has an excellent library of tools supported by a friendly Facebook group.
Something to get you started
You need to prompt generative AI to refine the results as it learns every time you prompt it. Open ChatGPT in your browser and start a free account. The app is available for phones too.
In the chat window, type: Summarise photosynthesis so I can teach it to 11-year-olds. When the results are in, type: include some questions to ask. Keep prompting it with things to include, or to make simpler, or include specific examples maybe to tie in with another area of the curriculum. Mine was ‘include an example from the Amazon rainforest’. Mind blown yet? We’ve only just started.
Curriculum planning
The information that LLMs have been trained on is from the internet. So, it has to be checked for accuracy. Errors are sometimes returned by the AI engine. I’ve challenged it a few times and it returns a message like: ‘Sorry, you’re absolutely correct, I’ll be more careful with my answers in the future’. I definitely prefer this to some human interactions I’ve had! This is why in curriculum planning, a subject leader needs to oversee the process, fact checking and refining it .
TeachMateAI has a number of tools available which have been trained on the National Curriculum and specific trusted sources. One of TeachMateAI’s main claims is that it takes the ‘heavy lifting’ out by saving time for teachers. The free version gives you access to a limited number of tools. The subscription option opens up all the tools. The beauty about TeachMateAI is that it’s been designed by teachers for teachers. You can download in various formats, there’s a lot of help for you to prompt the AI engine and it just looks lovely.
The Endpoint generator can enhance existing schemes of learning well, making sure the end points of learning are crystal clear. Then there is a slide show generator which builds a series of slides on any topic. It then generates questions on the topic. I can see this being really useful for supply or non-specialist staff. For schools who are starting from scratch with a scheme, a great place to start on TeachMateAI is the Medium term topic planner. This will return a sequenced series of lessons on any topic with objectives and activities.
School development, self-evaluation and quality assurance
The amount of time subject leaders put into development plans can be colossal. I tried the subject leader action plan tool, writing three weaknesses I recently saw in a visit to a maths department. TeachMate reports on how much time you saved with each tool. It took me 1 minute to write the weaknesses and it returned a full action plan with success criteria and actions. 0.75 of an hour saved, apparently. Some editing was needed of course, but the time taken on refinement was minimal.
A recent introduction to ChatGPT and other AI tools, is the ability to attach pdf files and paste text from documents or the internet. The recent subject surveys from Ofsted or DfE guidance can be lengthy and time-consuming to read in full. Generative AI will summarise documents in seconds. If you are development planning, AI will take into account Ofsted reports, lesson visit information and other quality assurance documents to produce a coherent development plan or self-evaluation document. I recently showed this to a principal who later told me she’d regained a week of her life.
Accessibility
Together with the many ways it can support SEND (something for another blog), improving accessibility is, in my opinion, the most significant contribution AI can make to education. A growing proportion of school leaders have some neurodiversity. Some may struggle with pitching the right tone in a letter to parents, or with writing up policies and other documentation. LLMs, because they have analysed so much language and contexts, are able to make language more or less formal. They can turn simple bullet points into long narratives. They can extract the key information from your writing and customize it for a particular audience and of course will check your punctuation, spelling and grammar!
Translation into other languages is nothing new of course, but with generative AI you can go deeper and well beyond what Google translate can manage. You can translate a document into Urdu, then prompt it to summarise, change the formality of the document, add subheadings…
Are our jobs safe?
My view is that AI is hugely positive to education in many ways. Workload is a constant struggle for teachers and AI can support a lot of the heavy lifting, giving time for teachers to spend with pupils. Nothing can replace the contact with one human being with another and supporting teachers with curriculum planning, quality assurance, absorbing official documents enables them to do just that.
Teachers with disabilities or neurodiversity will have the right tools to support them in the classroom and to progress to leadership positions. Currently only 13% of new entrants to teaching have a disability. Taking away barriers to accessibility through AI can improve this representation.
And moving forward?
Where generative AI is great, is helping with anything that is text based. That’s because LLMs analyse colossal amounts of words and phrases. Where AI isn’t fantastic at the moment is generating pictures or non-text-based characters, although that functionality is coming fast. A group of Year 12 students at the Leeds Maths School (LMaS), recently tried to find solutions to create teacher worksheets which includes the mathematical notation necessary at A level and beyond. Designing a worksheet with pictures is also tricky at the moment. AI will design some fantastic history questions but including pictures will have to be done manually for now.
Teacher awareness of AI is not great in this country at the moment. I’ve asked around. A lot. Occasionally I’ll get directed to the computer science teacher who is quite interested and is using AI tools for themselves. There’s a number of US based organisations making AI accessible to all teachers, supported by funded government programmes and courses. Over here, TeachMateAI is generating interest from teachers with a growing number sharing their discoveries on social media platforms. Particularly interesting is how TeachMateAI is developing AI Ambassadors in schools, sharing ideas and spreading awareness. It has now developed a ‘certified educator’ accreditation (which I got this morning!) and plans to roll out this programme at different levels.
In the absence of any centralised government interest in developing AI, it seems teachers themselves are beginning to drive the movement.
