ChatGPT and the application of A.I. in education: prompts and possibilities

New tech appears all the time and often disappears just as quickly in the world of education, with not many new things really becoming part of the fabric. We think you are right to be highly selective of where you invest your time, and we admire the fact that you don’t just fall in love with any old app. ChatGPT, however, is worth your interest.

So, to business. Let’s look at some practical ways to use ChatGPT in a learning environment. Why not try heading over to ChatGPT and pasting these prompts into the chat field to see for yourself what it produces.
How does AI look in education? Let’s ask ChatGPT

Prompt: Create a 20-question quiz to practice probability calculations for learners aged 14-17. The questions should get progressively more difficult. Give the answers at the end of the text.

Use: This creates a great series of problems for learners to collaborate on, and this can be created by the learners themselves rather than the learning guide, of course. They may have learned about probability in an experiential way, but exercises like this help to strengthen that mental model by letting them collaborate on varying angles to the problem and test out their ideas and consolidate them. When a gap is identified, challenges like these can be generated in seconds. This application of AI, like ChatGPT in education, allows educators and learners to generate anything from writing prompts to discussion questions, problems to solve, and so much more. Learners are able to generate these as and when they are most useful and self-direct their own exploration and practice.

Prompt: In the Latex software programme, how do I represent a differential equation?

Use: When learning to code, many hours are spent figuring out these mico-issues before you have a macro breakthrough. Until now, learners would google these questions, but as this Twitter thread makes clear, the way Google answers this question is nothing in comparison to the step-by-step explanation from ChatGPT, which includes visual examples and common pitfalls. In fact, these quick-fix explanations from ChatGPT are generally very well put together and support self-directed and independent learning.

Prompt: Write a creative short story about a Rottweiler called Huxley who discovered that he could speak to humans but was afraid to reveal his secret until he could trust the right person.

Use: Ok, that was weird, but bear with us. The thing is that OpenAI can definitely write this, but it won’t be elegant prose. Learners can really internalize the difference between machine and human output here by deciding how to improve the text. Style, tone, consistency, emotion, depth, complexity. This makes for great exercise.

Prompt: Explain the origins of the first world war, but insert three pieces of information that are incorrect. The text should be biased in favour of the austro-hungarian empire and written in an authoritative style.

Use: ChatGPT sounds like it’s confidently right but does certainly get things wrong. Humans, of course, get things wrong too, and sometimes quite intentionally. Information and media literacy is a core skill that the Finnish put a lot of stock in, and rightly so. Identifying bias and fact-checking information are core elements of critical thinking in a world where technology is increasingly able to muddy the waters, from deep fake videos to misinformation bots. Using tech to generate means to identify this is a form of poetic balance we rather like. Learners using this text to spot bias, how it is constructed, and how the false information is embedded is a great way to build these important skills and awareness.

Prompt: Ok, this is not a prompt, so don’t paste this into ChatGPT. The thing is that this tool can be used iteratively and developmentally too. We found one example where users were trying to generate ideas for an entrepreneurial side hustle, and we’d suggest trying it too and varying the business you choose. After each question, a response was generated, which made them prompt the tool to refine, develop and dig deeper. Here are the questions they asked:

How do you start a business? (general advice given)
Can you turn that information into a five-step plan? (a five-point checklist was given)
Can you suggest an online product or service I could offer? (a list of ideas provided)

At this point, one of the suggestions was selling handmade jewelry. The prompts continued:

How should I figure out what jewelry to make? (advice given on exploring and refining options and what factors go into that)

Things then got really interesting. ChatGPT provided advice on marketing and sales strategies, tools to help, how to create a website, and how to track progress and engagement.
Chat GPT is just getting warmed up

Why are we telling you this bit? Because it exemplifies where ChatGPT is going next. The current iteration is not connected to the internet. It only knows a little of the world beyond 2021, and so it cannot give you up-to-date market research. It writes you HTML and CSS code for a basic website but cannot create one for you. It can explain how your products should look, but it cannot visually represent this for you. Yet.

In 2025, ChatGPT web search will be able to carry out extensive market research for you online. It is totally within the scope of possibility that it can interface with other tools to build you that website. Using the amazing Dall-e open AI tool, it will be able to create that image of your handmade jewelry before it physically exists and show you how it looks while worn or displayed.

We are just scratching the surface here, but as educators, let us tell you that we are so excited at the potential of this to support learners in taking even greater control over their own learning and in bringing a new creative dimension to the learning environment that will facilitate boundless creativity.

Leave a Reply

Your email address will not be published. Required fields are marked *