EASIE AI - For learners, parents, and teachers [Issue 6]

Welcome, cutting-edge educators and aspirational parents

Introducing your no-nonsense EASie guide to using AI to support your kids’ learning and growing your own AI literacy

Today, we’ll cover:

• Trending AI Apps

• Using new tools - Claude Artifacts - Free

• Generating Images with AI - Letters and Themes

• AI Beginners Jargon—Hallucinations

• AI Adepts Jargon—Small Language Models

• Good prompting—Breaking it Down

• Hidden Dangers—Debunked theories reported as true

• AI News

We want to continue producing this free newsletter to help parents and teachers benefit from AI while keeping their kids safe. Please forward it to someone you think would benefit from its contents and help us keep it alive.

Reading Time 6 Mins

Trending AI Apps

AI apps delighting educators and parents.

• Dyslexia.AI Its early days for AI in SEN support but it has begun.

• AI Story Generator We don’t recommend allowing preteen kids unsupervised use of generative AI but you could use this app with them to start off their creative writing journeys.

• Merlin.AI lets you use most well-known AI models within a Chrome plugin. All advanced and premium features of Merlin are free to use.

Using New Tools - Claude’s Artifacts

Create online interactive quizzes

There is more to AI than ChatGPT and Claude is a competitor of simialr capability and an interesting ‘artifacts’ feature.

Artifacts are standalone content created by Claude such as an interactive quiz. Here’s how to create an interactive quiz in Claude (free version).

Step-by-step

• Sign in on https://claude.ai/

• Activate ‘artifacts’ as an experimental feature.

• Drag a document such as a PDF featuring your source text into the window.

• Prompt Claude to create an interactive learning game using Artifacts.

This is the prompt I used. It will work with any text. You could save a Wikipedia article as a pdf or get a short story from Project Gutenberg as I did.

Prompt: Use artifacts to create an interactive game for learners to practice their reading comprehension skills based on the attached file. You should always give an explanation after each question explaining why the answer is correct. Show the relevant part of the text for extra clarity. The quiz is aimed at learners with grade 6 level reading.

Teaching/Learning Use

Claude is not for unsupervised use by preteens but you could use this workflow to create quick supervised tests.

Generating Images With AI

Fun with materials

Use any image generator and ask it to generate the object in a specific material, even if it makes no sense in reality.

How does this help in education?

It's a thematic way to put a title on your kids’ display work. For example, their work on volcanos could ask for text made of lava.

But that’s not the learning I have in mind here.

I’m thinking of this to get kids to reflect on the keywords they associate with a topic. You could work on these with a class or discuss what the links are such as helping them remember the name and role of the River Nile.

An example will help.

If you’ve studied Ancient Egypt, work one letter at a time and put in a prompt to make that letter represent an aspect of the topic.

For my Egypt text, I used:

• The letter E made of sand

• The letter G formed by a river in a desert

• The letter Y printed on a papyrus sheet

• The letter P wrapped in sepia coloured bandages

• An Egyptian ankh stood in the sand of a desert

(I combined them in a separate app - you can use whichever art package you prefer or even in a Word doc).

Sometimes you’ll need to give a bit of extra help, such as using sepia to make the bandages more mummy like rather than medical. For the Ankh, I just needed it to be thematic in its surroundings.

AI Jargon - Learning

We’ll help you increase your knowledge of AI by explaining a beginner and advanced concept each week.

For AI beginners—We all start somewhere.

Hallucinations

In AI, a hallucination refers to a scenario where the AI generates information or content that is incorrect, nonsensical, or fabricated. These hallucinations can happen in various types of AI models.

They can arise for different reasons but common ones are that the AI doesn’t include the relevant information in its training or the user’s prompt did not give enough clarity.

The AI gives hallucinated answers with as much confidence as correct answers which can mislead learners. Learning not to rely on AI but to fact check it is a key part of AI literacy with the current level of the technology.

For AI Adepts—Next-level AI Knowledge!

In AI, SLM (Small Language Model) and LLM (Large Language Model) refer to the size and capacity of the language models.

Small Language Models contains fewer parameters (the connections within the AI’s knowledge) than LLMs. They are generally less accurate and capable but faster and requires less computational resources.

SLMs can be very effective when trained to perform dedicated tasks and we can expect to see them running on mobile devices soon. As they don’t need to connect LLMs hosted by third parties, SLMs on a device can be much more data secure.

Prompt Guide

This week's Prompt Better example could also be called ‘How to think better’, as it is a good approach even for humans.

Think of the saying, "How do you eat an elephant? One bite at a time"* 🐘

Essentially, take any huge task or complex problem and break it down into simple steps or tasks.

Let's consider the following question: Could King Henry VIII have met Shakespeare? They both lived during the period of English history, often called the Tudor times, so it seems likely. 

However, the Tudor period was over a century, so maybe not.

If we ask an AI this question, it can go wrong as it jumps to conclusions based on incomplete information. Humans do the same.

However, if we tell it to answer step-by-step, we have a much better chance of getting the correct answer, and we can easily check the reasoning and foundational facts. 
This is a very simple prompting technique, considering its influence on the output.

Prompt: Was it possible for King Henry VIII to have met Shakespeare? Think step by step.
Output: To assess whether King Henry VIII could have met William Shakespeare, let's go through the timelines of both individuals step by step:

King Henry VIII's Lifetime: Henry VIII was born on June 28, 1491, and he died on January 28, 1547. His reign as King of England spanned from April 21, 1509, until his death.

William Shakespeare's Lifetime: Shakespeare was born on April 23, 1564, and died on April 23, 1616. He was an English playwright, poet, and actor, considered one of the greatest writers in the English language and the world's greatest dramatist.

Given these dates, King Henry VIII died in 1547, which is 17 years before William Shakespeare was born in 1564. Therefore, it would have been impossible for King Henry VIII to have met William Shakespeare, as they lived during different times, with a gap of several decades between the death of Henry VIII and the birth of Shakespeare.

This prompting technique is obviously good for improving the quality of your AI-generated outputs, but you can also use it to teach kids.

Give your child the same prompt and ask them to show their thinking step-by-step. Let them access their usual sources of information, such as books or web searches, but not AI. They should show their reasoning in simple, clear steps. Compare their answer with the AI prompts and resolve any discrepancies. Remember, AI is not infallible, so don't assume it is right and your child is wrong.

This exercise will have two desirable learning outcomes. Learners will recognise the importance of not feeling overwhelmed by a complex task and instead breaking it down. They will also learn about AI and how it works, which will be an ever-growing need in the future.

*Apologies to vegetarians and vegans.

AI Hidden Dangers

To have good AI literacy means you must be aware of AI’s limitations as it is at the moment. One of these limitations is that it reflects its training material.

This is fine if the material is accurate but what was once thought to be true has later been discovered false. Often the old information fades as the new understanding takes over but sometimes old beliefs held such sway they persist.

Discussed and spoken about in internet chats and publications they leave a heavy mark on AI’s training. There are many debunked theories of education you’ll see mentioned with confidence by AI. See our article to see what to look out for.

Find out more in our Teachers using AI - a hidden danger

AI News

Your one-stop, in-brief AI newsstand

• Two-thirds of high school and college teachers say they use AI technology for education. Quizlet/Newswire

• Apple’s introduction to AI looks likely to be later than the hoped-for September as it does not feature in the current beta release for cutting edge users.

• AI is getting harder to avoid as Amazon released Rufus for all US customers. Rufus is Amazon’s AI-Powered shopping assistant.