Y9 How AI Works

Back to:

AI Art Generation

Next Topic

Prompt Engineering

A really smart machine!

This is a google AI Experiment where their AI has been trained on the worlds largest dataset of doodles.

Click- "Let's Draw" and you get 20 seconds to draw something and the AI will try to guess.

Work with a partner and, as you draw, see who can guess first; your partner (a human) or the machine!!!

I think you'll be very surprised.

OR

The Intelligent Sheet of Paper

In pairs you are going to play Noughts and Crosses. One of you is going to be play normally and one of you are only going to do what the piece of paper says.

Swap over each time and play a few rounds.

You can also see a Chicken Play.

Machine Learning

How Machine Learning Works*

*This is a simplified explanation of how AI works.

Artificial Intelligence is a complex and rapidly changing field. For this lesson, we are only looking at the basic ideas behind machine learning and how Large Language Models recognise patterns in data.

If you are interested there are more detailed explanations in the Level 2 Computer Science.

Your Turn!

Kiwi, Kiwi, or Kiwi?

Kiwis are confusing so we are going to make a Machine Learning model to tell them apart.

Task 1: Introduction to Machine Learning

Maungatautari Ecological Island Trust, Public domain, via Wikimedia Commons

By Ivar Leidus - Own work, CC BY-SA 4.0, https://commons.wikimedia.org/w/index.php?curid=116097447

By Keith Allison from Hanover, MD, USA - Steven Adams, CC BY-SA 2.0, https://commons.wikimedia.org/w/index.php?curid=56273728

OR

Teachable Machine

Train your own AI

This is another Google AI experiment. With a webcam you can train an AI using "Machine Learning" to recognize image (faces, items etc) or sounds (like two a click or a clap).

Get a partner and try it out!

Can you train it to tell the difference between both of your faces?

Large Language Models

Large Language Models (LLMs) like ChatGPT and Gemini are computer programs that generate text by predicting one word at a time. They are trained on large amounts of text written by people and learn patterns about which words usually come next. When you type a prompt, the model looks at the words so far and chooses a word that is likely to follow. It repeats this process again and again to create sentences. LLMs do not understand meaning like humans do, they recognise patterns in language and use those patterns to produce text that sounds natural.

Task 2: Introduction to Large Learning Models

Open the LLM Worksheet

In this worksheet, you are acting like a Large Language Model. You are choosing one word at a time based on what could reasonably come next, rather than planning the whole sentence in advance.

As a class make a one word story to see how you can add words to a sentence or story.

Fast finishers

Contexto is a word-guessing game where your guesses are ranked by how similar they are in meaning to a hidden word. This is similar to how Large Language Models work, as they learn patterns and relationships between words based on how they appear together in large amounts of text.

Page updated

Report abuse