In this activity, students delve into affective computing by creating an expressive robot face out of materials and programming it to react to various sounds with different emotions, and later come up with new facial expressions and corresponding rules.

http://www.cs4fn.org/teachers/activities/createaface/createaface.pdf

In this activity, participants will work with treasure maps, which operate similarly to finite-state automata used by computer scientists to process sequences of symbols like words or letters in a document or computer program.

https://classic.csunplugged.org/documents/activities/finite-state-automata/unplugged-11-finite_state_automata.pdf

https://classic.csunplugged.org/documents/activities/finite-state-automata/unplugged-11-finite_state_automata.pdf

AutoDraw is an application developed by Google that uses machine learning to help users draw simple sketches. The user starts drawing a rough image, and the AI system suggests a set of possible matches, which the user can select from to improve the image. The tool can be used for creating art, diagrams, charts, and other visual aids, and its suggested images come from a large dataset of user-generated images and icons.

How it works: https://experiments.withgoogle.com/autodraw
Try it: https://www.autodraw.com/

"Handwriting with a Neural Net" is a Google experiment that uses machine learning to recognize and generate handwriting. Users can write any word or sentence on a touchscreen or trackpad, and the system will generate similar-looking handwriting in real-time. The system is based on a neural network trained on a large dataset of handwriting samples and can generate a variety of handwriting styles. The experiment demonstrates the potential of machine learning in improving the accuracy and efficiency of handwriting recognition and generation.

See the Experiment: https://distill.pub/2016/handwriting/

“Teachable Snake is an interactive web game that uses the beta version of Teachable Machine 2 and React.js,” where players can draw a black arrow on a piece of paper to control the movement of the snake using machine learning, without needing physical buttons.

https://experiments.withgoogle.com/teachable-snake

The Teachable Machine is a Google experiment that allows users to train their own machine learning models without any coding knowledge. By using their device camera and microphone, users can teach the machine to recognize and classify different objects or sounds based on their own examples, creating custom machine learning models that can be exported for use in other projects or applications. This experiment demonstrates the accessibility and democratization of machine learning technology, empowering users to explore the possibilities of artificial intelligence in a user-friendly and interactive way.

Information: https://experiments.withgoogle.com/teachable-machine
The Teachable Machine: https://teachablemachine.withgoogle.com/
The Teachable Machine Version 1 (less powerful but easier to use): https://teachablemachine.withgoogle.com/v1/

In this activity, students play Hexapawn, a chess like game, against a machine made of sweets that starts with only basic knowledge of the game and improves as it learns from its mistakes, with the class punishing it by eating its sweets when it loses.

http://www.cs4fn.org/teachers/activities/sweetcomputer/sweetcomputer.pdf

Quick, Draw! is a Google game that challenges players to draw objects within a limited time frame while an AI tries to guess what they are drawing based on its previous training data. The AI uses machine learning algorithms to analyze the players' sketches and make guesses about what they represent, providing feedback to improve its accuracy over time.

Link to game: https://quickdraw.withgoogle.com/
Link to Data: https://quickdraw.withgoogle.com/data

Sketch RNN works by learning the patterns and styles of hand-drawn sketches in a large dataset, and then generating new sketches based on that knowledge. To create a sketch, the user starts by drawing a rough outline of an object, and Sketch RNN generates a sequence of strokes that progressively refine the drawing. The system uses a combination of a variational autoencoder and a recurrent neural network to predict the next stroke in the sequence based on the previous ones.
How it works: https://magenta.tensorflow.org/sketch-rnn-demo
Try it: https://magenta.tensorflow.org/assets/sketch_rnn_demo/index.html

Tiny Sorter is a DIY experiment that uses Arduino and Teachable Machine to teach physical computing and machine learning in a simple and fun way, by building a machine that can sort objects using a machine learning model, which can be created easily without coding using Teachable Machine. It requires an Arduino, a servo motor and a few crafting materials. With a bit of experience, it could also be adapted to work with the BBC Microbit.

https://experiments.withgoogle.com/tiny-sorter

Emoji Scavenger Hunt is a fun and engaging mobile game that combines the excitement of scavenger hunting with the novelty of emojis. The game uses machine learning technology to recognize objects which creates an innovative and unique element to the gameplay.

Overview: https://experiments.withgoogle.com/emoji-scavenger
Play the game: https://emojiscavengerhunt.withgoogle.com/

“This site will help bridge the gap between the Teachable Machine AI and a micro:bit giving you clever new ways to shape your projects. Train an AI to make a prediction using a library of data you give it, and then code your micro bit to use those predictions to activate motors, lights, and more! Simply click on "Pair Microbit" and follow the steps to get started today!” (ai-training, n.d.).

https://ai-training.glitch.me/
Demonstration Video: https://www.youtube.com/watch?v=Lx0l5U-vVk4