Beginner Tutorials - The Human in the Room

1.1 How to Get Started

The very first thing we need to do if we want to use generative AI is to create an account with one of the companies who have developed large language models. The most popular ones are, in no particular order: ChatGPT by OpenAI, Google's Gemini, Claude by Anthropic, and Copilot by Microsoft. On your browser, type any of the names listed above along with the words 'Create an account'. This will take you to the relevant website, where you will simply have to follow the instructions: enter your name, your email address, and, in some instances, your mobile phone number.

For those who are not comfortable with creating an account just yet and would like to try out one of the chatbots, ChatGPT and Copilot offer a limited version of their models without the need of registering with them. I recommend Copilot for this, since ChatGPT only allows you to ask a few questions before prompting you to create an account.

It's always worth having a look at the privacy policies. But one of the key ideas you need to be aware of is that, unless specified otherwise, or unless you activate a feature that prevents this, all the data you enter and all your conversations with the chatbot can, and most likely will, be used to train future models. It's always a good idea to be mindful of this, especially when it comes to students' data, which must not, under any circumstances, be entered into a large language model with unrestricted access to your conversations.

Once you've got your account ready and you have signed in, you can start asking the chatbot to generate whatsoever pedagogical material you are interested in. As you progress in this chapter, you will find some amazing uses along with a few limitations, or things that the model can't simply do.

The interfaces work very much in the same way a chat application does, like WhatsApp or Messenger. You simply enter your text in the chatbox and press enter. That will prompt the large language model to address your question. And it very much works as a dialogue, where you can iterate and refine as you wish. If you are a total beginner, I recommend that you start by having a chat as if you were talking to an actual person. You will be impressed by how natural the discourse can be.

1.2 Questions

For practical reasons, I will assume that you have used web searchers like Google before. It is precisely in the case of Google and its chatbot, Gemini, that the frontier between a web searcher and a chatbot are particularly blurry. You can, therefore, type questions in the style you were used to typing when you were searching something online. The crucial difference is that you will get collated results and explanations of the sources searched, instead of having to browse the results of your search to try to find something relevant.

There are, however, a couple of elements that I want you to consider. When you do a search or you ask a question to a large language model, the first one is that, similarly to a human being, it will fill up the gaps in your question, meaning that if there's not enough context, the large language model will imply what you mean. For example: 'What's a good activity for my pupils?' Some large language models will challenge questions like this and will ask you for context, but they may still attempt to produce an answer. Will said answer be relevant to you? Probably not, because the AI model will need to make lots of inferences and assumptions as of: what type of activity do you want?; the age and level of your pupils?; the purpose of the activity?; the subject?; the topic?; the language?; the example?; the syllabus?; the part of the lesson?, etc. Specificity, therefore, is key. In the 'Prompt Engineering' section, I review the most important elements that need to be present in your instructions or questions to a chatbot.

The second element that distinguishes large language models from web searchers is the ability to iterate and to refine your questions. You don't need to be content anymore with the poor results of a web search. You can now ask the large language model to provide a more refined version of the results, or the answer to your question, until you get exactly the response you need. An example of a good question would be: 'What’s a good starter activity for my year 11 GCSE biology pupils who have just seen an introduction to photosynthesis, so I can get them started for a lesson where the objectives are to consolidate the understanding of the concept of photosynthesis and all the parts involved in the process?' If you prompt like this, you will get a few suggestions, but you can always iterate so the large language model refines the answer. For example: 'I like the idea of the game. Can you give me three more ideas?' In this second question, you won't need to provide further context because the large language model will remember and be able to recall all the conversation and all the previous interactions you have had with it within the same chat.

LLM’s response:

A great starter activity for your Year 11 GCSE Biology pupils — especially just after an introduction to photosynthesis — should be engaging, quick to set up, and help them recall and apply prior learning. Here's one that works well:

Starter Activity: "Photosynthesis Dominoes"

Time: 10 minutes

Objective: To match key terms with their definitions or roles in the process of photosynthesis.

🧩 How It Works:

Prepare a set of domino-style cards where each card has a term on one side and a definition or related concept on the other.
Students work in pairs or small groups to match the terms to the correct definitions, creating a full chain.
The last card should link back to the first if possible, making a loop (optional but satisfying!).

🧪 Key Terms to Include:

Photosynthesis, Chlorophyll, Chloroplast, Carbon dioxide, Water, Glucose, Oxygen, Sunlight/light energy, Stomata, Xylem, Palisade mesophyll, Word equation, Balanced symbol equation (optional for higher ability).

🧠 Extension / Differentiation:

Higher ability: include symbol equations or more complex components like NADP, light-dependent/independent reactions.
Lower ability: focus on the word equation and key organelles only.

✅ Why It Works:

It reinforces key vocabulary and conceptual links.
It encourages peer discussion and collaboration.
You can walk around and assess understanding informally.

Would you like a printable template or digital version of this to use right away?

1.3 Texts

Large language models can generate texts in any style or tone, for any subject and topic, and with any characteristics you specify. This is a game-changer for teachers of those subjects in which reading and understanding is key, since we are now able to tailor the content to the specific needs and abilities of our classes. We have, again, to be as specific as possible. And one of the examples I use in the 'Prompt Engineering' section is precisely with text crafting.

One piece of advice is that you check always the large language model output to verify all the content there is true and correct, and there aren't any instances of hallucinations. To see what 'hallucination' is in the context of large language models, please refer to the relevant section in this book.

Sample prompt:

“Act as a historian specialising in modern European history. Create an informative essay of approximately 500 words. The text should be written in a formal and expository style. The target audience is Year 11 history students. The topic of the text is the main causes of World War I. It should cover the following key points: The system of alliances (Triple Alliance and Triple Entente), Militarism and the arms race, Imperialism and colonial rivalries. The main objective of this text is to explain the long-term factors that contributed to the outbreak of World War I. Please structure the text into 4 paragraphs and include a list of challenging vocabulary with definitions at the end.”

LLM’s response:

The Main Causes of World War I

The outbreak of World War I in 1914 was the result of a complex web of long-term factors that had been building over several decades. Chief among these were the system of alliances, militarism and the arms race, and imperialism with its associated colonial rivalries. Together, these elements created a highly volatile political landscape in Europe, in which even a small incident could escalate into a full-scale war. Although the immediate cause of the war was the assassination of Archduke Franz Ferdinand of Austria-Hungary, it was these underlying tensions that made a general conflict almost inevitable.

One of the most significant long-term causes of the war was the system of alliances that divided Europe into two hostile camps. On one side stood the Triple Alliance, formed in 1882, which united Germany, Austria-Hungary, and Italy in a defensive pact. In response, the Triple Entente was established by 1907, comprising France, Russia, and Britain. While these alliances were intended to provide security and balance, they had the opposite effect by increasing suspicion and encouraging nations to rely on military force. The alliances also meant that any conflict involving one country could quickly draw in others, turning a regional dispute into a continent-wide war.

Closely linked to the alliance system was the growth of militarism and the arms race. Militarism is the belief that a country should maintain strong armed forces and be ready to use them to defend or promote national interests. In the years leading up to 1914, the major European powers dramatically increased the size and strength of their armies and navies. Germany, in particular, sought to challenge British naval dominance by expanding its fleet, prompting Britain to respond in kind. This arms race created an atmosphere of fear and competition, where military planning and war strategies became central to national policy, and generals held significant political influence.

Imperialism and colonial rivalries also played a crucial role in building tensions between the great powers. European nations were competing fiercely for overseas territories, particularly in Africa and Asia. These rivalries often led to diplomatic crises and increased hostility between countries. For example, the Moroccan Crises of 1905 and 1911 saw Germany challenge French control of Morocco, straining relations between Germany, France, and Britain. Imperial ambitions not only heightened competition but also contributed to national pride and a willingness to assert dominance, even at the risk of war.

In conclusion, while the assassination of Archduke Franz Ferdinand was the immediate spark, the roots of World War I lay in deeper structural issues. The rigid alliance system, rising militarism and the arms race, and aggressive imperial competition created an environment primed for conflict. These long-term causes meant that Europe was a powder keg, and once the fuse was lit, the continent descended into one of the deadliest wars in human history.

Challenging Vocabulary:

Alliance – A formal agreement between two or more nations for mutual support in case of war.
Militarism – The belief in building up strong armed forces and being prepared to use them.
Imperialism – A policy of extending a country’s power and influence through colonisation or military force.
Colonial rivalry – Competition between countries to acquire and control colonies, especially in Africa and Asia.
Diplomatic crisis – A serious conflict between countries that can lead to war if not resolved.
Volatile – Likely to change rapidly and unpredictably, especially for the worse.
Assassination – The deliberate killing of a prominent or important person, often for political reasons.
Dominate – To have control or influence over others.

1.4 Summaries

Have you ever felt there's a lot of very interesting articles and books you would like to read, but you don't really have time to do it? Now, you can use AI as your time machine by asking it to summarise any text of any length. Whether you're focussing on your professional development, research, or simply because you have found a lengthy article with a catchy title in your social media feed, you can use AI not only to summarise the content but also to get said content organised in the way you desire. You can ask a large language model to list the key takeaways of an article or book, to explain the core idea, to summarise chapter by chapter, and to present the content to you in bullet points, short paragraphs, or one long paragraph, highlighting with italics, bold, underlined words, and even emojis to offer visual support.

This can be done with any language model, but in the tools chapter of this book, we see how NotebookLM, powered by Google, is one of the best tools to summarise your texts. Summaries are an excellent learning tool, and using AI can smooth the process of preparing learning guides, revision booklets, and other types of learning materials that can help your students digest the content of your subject. A summary prompt can range from the most simple, "Summarise this text," to the more complex:

Sample prompt:

"Act as a teacher of geography. I need you to summarise this text about tectonic plates for my Year 11 class to understand the key concepts. Present the content in bullet points in a clear and explicative style that is easy to understand for 15-year-olds."

LLM’s response:

Here’s a simplified and clear bullet-point summary of the tectonic plates topic, designed for your Year 11 geography class (around age 15):

🌍 The Restless Earth: Key Concepts About Tectonic Plates

The Earth's surface is not one solid piece – it’s made up of huge sections called tectonic plates.
These plates are part of the lithosphere, which includes the Earth's crust and upper mantle.
Tectonic plates are constantly moving, very slowly, due to heat and pressure from the Earth’s core.

🔥 Why Tectonic Plates Matter

This movement causes major geological events like: Earthquakes, Volcanic eruptions, Mountain formation. It also helps create new land and recycles old crust – it's part of the Earth's natural cycle.

🧭 Plate Boundaries – Where the Action Happens

There are three main types of plate boundaries, where plates meet and interact:

Divergent Boundaries – Plates move apart: Magma rises up, cools, and creates new crust. Found in places like mid-ocean ridges (underwater mountains). This process is called seafloor spreading.
Convergent Boundaries – Plates collide: One plate may slide under another – this is called subduction. Can form deep ocean trenches, volcanoes, and mountain ranges. When two continental plates crash, they crumple and rise to form mountains (e.g. Himalayas).
Transform Boundaries – Plates slide past each other: This causes fault lines. The stress builds up and is released as earthquakes. A famous example: the San Andreas Fault in California.

🌋 Tectonic Activity and Natural Hazards

Earthquakes are caused by sudden release of pressure along fault lines. Volcanoes often form at subduction zones when melting rock (magma) rises. These events can be dangerous but are also important for the planet’s balance: they create new land and release gases that help shape the atmosphere.

🌐 Why This Matters

The theory of plate tectonics helps explain how continents and oceans are arranged, why earthquakes and volcanoes happen where they do, and how Earth has changed over millions of years. Scientists use this knowledge to predict natural disasters and help keep people safe.

1.5 Adapting Texts

In a similar way to summaries, you can feed your large language model with a text of any nature and ask it to adjust the level. You can either simplify it to make it easier to understand, or you can make it more complex by asking the large language model to add specific concepts, structures, or vocabulary that your pupils need to learn. Once again, the idea is quite simple, and your prompt can go from a mere, "Adapt the reading level of this text to my 14-year-old class," to more complex or elaborated prompts like:

Sample prompt:

"Act as a science education specialist. Adapt the reading level of this text to my 12-year-old pupils so it is age-appropriate and engaging, and to help them the physics of light — waves particles and colour. Rather than avoiding complex concepts, use strategies like analogies, metaphors, storytelling, and simple explanations to explain them. Provide the necessary scaffolding for my 12-year-old pupils to understand the core ideas.”

LLM’s response:

Certainly! Here's an age-appropriate and engaging adaptation of the text for your 12-year-old pupils, using analogies, metaphors, and simple but not oversimplified explanations. The goal is to scaffold understanding of the physics of light, including waves, particles, and colour, while keeping the wonder of the topic alive:

The Strange and Amazing Nature of Light: Waves, Particles, and Colour

Imagine light as a superhero with two amazing powers — it can travel like a wave, and also act like a tiny particle called a photon. That’s right! Light doesn’t choose just one identity — it can be both at the same time. This strange and wonderful idea is called the dual nature of light, and it’s one of the coolest mysteries in science.

🌊 Light as a Wave: The Wiggly Traveller

Let’s start with the wave side. You’ve seen waves at the beach, right? Light waves are similar, but instead of water going up and down, light has invisible electric and magnetic fields that wiggle as it zooms through space at the fastest speed in the universe — about 300 million metres per second! These wiggles go sideways, like someone waving a skipping rope up and down while running forward — this kind of wave is called a transverse wave.

Light waves come in different sizes, called wavelengths. Some are super short, and some are longer. The light we can see with our eyes (called visible light) is just a small part of a much bigger family called the electromagnetic spectrum. Visible light waves range from about 400 nanometres (violet) to 700 nanometres (red). (A nanometre is a billionth of a metre — super tiny!)

⚛️ Light as a Particle: The Photon Side

Now here’s where things get wild. Even though light travels like a wave, it sometimes acts like it’s made of tiny packets of energy — like drops of water from a hose, but way, way smaller. These energy packets are called photons. In the early 1900s, Albert Einstein discovered something weird: when light shines on certain metals, it can kick out tiny particles called electrons. But only light with enough energy can do this — like violet or blue light, not red. This mystery couldn’t be explained by just thinking of light as a wave.

Einstein’s idea? Light must be delivering energy in chunks — photons! The more energy a photon has, the better it can knock out an electron. Think of photons like soccer balls. A soft kick (a red-light photon) won’t move the ball (electron), but a strong kick (a violet-light photon) will send it flying!

🌈 How We See Colour: The Light Show in Your Eyes

Have you ever wondered why a strawberry looks red and grass looks green? When white light (which has all colours mixed together) hits an object, some of the colours are absorbed and others are reflected. The ones that bounce off are the ones we see. So if something reflects red light and absorbs the rest, it looks red to our eyes. Inside your eyes are special cells called cones — not the ice cream kind! These cones come in three types, and each one is good at picking up certain colours of light: red, green, or blue. When light enters your eye, these cones send signals to your brain, which figures out what colour you're looking at. So the colour of things isn’t really “in” the object — it’s how our eyes and brain react to the light bouncing off it!

🔬 The Mystery Continues...

Even though we’ve learned a lot, scientists are still exploring how light works, especially in quantum physics — the study of the super tiny world of atoms and particles. At that level, light behaves in surprising ways we’re still trying to fully understand. It’s like light is playing a cosmic game of hide-and-seek, showing us just a little bit of its secrets at a time.

1.6 Vocabulary Lists

One of the most useful ideas I have found to leverage AI is to use it to create glossaries or lists of vocabulary to help my pupils understand the keywords relevant to my subject. You can either ask for a vocabulary list on a specific topic, or you can feed the large language model with a text and ask it to create a glossary with the key concepts. This is one of the most relevant tools to support literacy and avoid our pupils' performance being hindered by them not understanding some of the key concepts of your subject when tackling exam-style questions.

Again, your prompt can range from the most simple question to a more complex and elaborated set of instructions, but you always have to keep in mind that if your question is vague, the large language model output will also be vague; if your question is specific, the large language model output will be specific too, to match your request. But as ever, there's nothing to worry about if the output you get from the large language model doesn't match your expectations. You can always iterate, and this is something you almost certainly will have to do. In fact, getting less-than-perfect answers and iterating whatever number of times will help you understand how a large language model works and when and how they need you to be more specific about a particular aspect. So please, do not smash your computer if you ask for a list of 50 words and you get 48!

Sample prompt:

“Act as a teacher of Spanish. I need you to generate a vocabulary list for my Year 11 class who will sit their AQA GCSE exam this year. Topic: environment and awareness. Ensure there are both low and high frequency words suitable for this level. Present your output in a table, with the Spanish words on the left column and the English translation on the right.”

LLM’s response:

Here’s a vocabulary list tailored for AQA GCSE Spanish, Topic: Environment and Awareness — including a mix of high-frequency essentials and some more ambitious words to stretch Year 11 students.

Spanish	English
el medio ambiente	the environment
la contaminación	pollution
el cambio climático	climate change
el calentamiento global	global warming
la deforestación	deforestation
la sequía	drought
el reciclaje	recycling
los residuos	waste / rubbish
la energía renovable	renewable energy
ahorrar energía	to save energy
desperdiciar	to waste
la huella de carbono	carbon footprint
los gases de efecto invernadero	greenhouse gases
la capa de ozono	the ozone layer
concienciar	to raise awareness
perjudicial	harmful
las especies en peligro de extinción	endangered species
proteger	to protect

1.7 Brainstorming

If you have never been stuck thinking about a type of activity or a game to improve one of your old lessons, it's probably because you haven't been in the profession long enough. I've lost count of how many times I finished planning a lesson just to think, "I'm gonna bore my pupils to death," and many times it's not very obvious what type of dynamic activities or games you can introduce to make your lesson more fun.

Simply ask the large language model to give you a few ideas for a game, a kinaesthetic activity, or a fun starter that you can use to make your lesson more fun. You will see that some of the ideas will be rubbish, but some others will be very good, and even things you would never think about, since some of the most advanced models can use web search to find materials from other people all over the world. In a similar way to the other examples I have been discussing, the more specific you are when you write your request to the chatbot, the more specific its output will be. So, indicating the purpose of the activity or game, where in the lesson sequence it will be placed, the target audience, the subject, the topic, etc., is important.

Sample prompt:

"Help me brainstorm ideas for a starter game for my Year Seven class who are learning how to structure paragraphs in English.”

LLM’s response:

Got it — Year Seven, English, paragraph structure. We’ll want something engaging enough to wake up even the most determined daydreamers, but simple enough that it reinforces the point (topic sentence, evidence, explanation) without getting buried in complexity.

Here are some starter game ideas:

1. Paragraph Relay

Setup: Split the class into small teams.

How it works: You give each team a topic (e.g., “School uniforms,” “The best season of the year”). Student 1 writes the topic sentence. They pass it to Student 2, who adds evidence (fact, example, or detail). Student 3 writes the explanation. First team to finish with a logical, complete paragraph gets a point.

Why it works: Fast-paced, collaborative, and forces students to see paragraph structure as a sequence.

2. Paragraph Jenga

Setup: Write parts of different paragraphs (topic sentences, evidence, explanations, random extra fluff) on Jenga blocks or slips of paper.

How it works: Students pull a block and have to decide if it’s a topic sentence, evidence, or explanation, then slot it into a “paragraph tower” in the correct order. Twist: Throw in a few “off-topic” sentences they must reject.

Why it works: Adds a kinesthetic, puzzle-solving element.

3. The Mystery Paragraph

Setup: Give them a scrambled paragraph (all sentences out of order).

How it works: In pairs or small groups, they race to arrange the sentences into a logical, well-structured paragraph. Bonus round: Ask them to justify why they chose that order.

Why it works: Forces them to recognise logical flow and cohesion.

If you’d like, I can put together one fully scripted starter game that’s ready to drop into your lesson plan — including topic ideas, instructions, and scoring — so you can run it straight away without extra prep. That way it’s more plug-and-play for your Year Seven group.