👋 Welcome to our World of AI lesson!
In this lesson, we will learn the various types of AI and applications, getting hands-on experience with some remarkable technologies!
So far at camp, we’ve dived into some foundational concepts that exist in many AI/ML technologies. In the KWK experience, we think it’s important to get a breadth of exposure to technology so that after camps, you can continue to pursue what you’re passionate about!
Introduction
After applying machine learning to natural language processing tasks like building a chatbot, you've gained a solid foundation in understanding AI. We've explored some amazing topics in the field of Artificial Intelligence, including sentiment analysis, neural networks, and deep learning, but there's still so much more to explore!
- Computer vision allows machines to interpret visual data, powering advancements in autonomous vehicles.
- Reinforcement learning develops systems that learn from their own experiences, informing how robots navigate complex environments.
- Expert systems mimic human decision-making in specific domains like healthcare, where they support healthcare professionals with disease diagnosis and treatment suggestions.
Let's take the next step in our journey as we explore these diverse subfields and see how they interconnect to form the broader AI landscape.
👩💻 Exploration #1: Play Pictionary with a Neural Network
Explore: start by exploring the website to the left, Quick, Draw! Write down what you think may be happening!
Here’s a bit about the AI behind Quick, Draw!: You interacted with a specific type of neural network called a convolutional neural network (CNN). A convolutional neural network (CNN) uses specialized layers called convolutional layers that apply filters to detect local patterns like edges, shapes, and patterns in data, particularly images. Unlike traditional neural networks, where every neuron connects to every neuron in the next layer, CNNs use shared weights and local connections, making them much more efficient for processing spatial data.
Think of a CNN like a detective with a magnifying glass looking at a photo! A traditional neural network would try to memorize every single pixel in the entire picture all at once, like trying to remember a whole page by staring at it. But a CNN works more like how your eyes actually work - it uses its "magnifying glass" (the convolutional filter) to scan small sections of the image at a time, looking for specific clues like "Is there a curve here?" or "Is this a straight line?" As it moves across the image, it builds up an understanding by finding patterns, first simple ones like edges, then more complex ones like "this looks like a wheel" or "this might be a face.”
Write a definition of CNNs in your own words!
👩💻 Exploration #2: Webcam Pac-Man Controller
Webcam Pac-Man Controller
https://storage.googleapis.com/tfjs-examples/webcam-transfer-learning/dist/index.html
https://storage.googleapis.com/tfjs-examples/webcam-transfer-learning/dist/index.html
Explore: Start by exploring the website above, Webcam Pac-Man Controller experiment! (You will have to grant access to your camera) Write down what steps may be happening behind the scenes!
Here’s a bit about the AI behind Webcam Pac-Man Controller: The AI behind this experiment relies on a machine-learning model powered by TensorFlow, an open-source AI framework. You are the one who trains the model — by showing your webcam examples of each hand pose, you build the training data yourself. Through this process, the model learns to associate each pose with a class or label, such as "left," "right," "up," or "down," just like you did when training your own classifier in Teachable Machine.
During the actual game, the AI model analyzes the live feed from your webcam in real time, classifying each hand pose into one of the directions you taught it. It uses computer vision techniques to identify and distinguish between poses based on the examples it was shown.
Computer vision is a field of technology where computers learn to understand and interpret visual information, just like our eyes and brains. It involves teaching computers to 'see' images or videos, recognize objects or faces, and make sense of the visual world around us.
The model applies techniques such as image processing, feature extraction, and pattern recognition to interpret your hand gestures in real time. It compares what the webcam sees against its learned knowledge of each pose, enabling it to make a prediction and send the correct direction command to Pac-Man, turning your movements into controls.
Beyond gaming, this kind of technology has real-world applications in accessibility. For people who cannot use a traditional keyboard or mouse, computer vision can turn body movements into inputs, making technology usable for everyone. Tools like eye-tracking software and gesture-controlled interfaces already exist to help people navigate devices using just their eyes or hands. The same ideas powering your Pac-Man controller are at the heart of these assistive technologies!
Beyond games, where else could gesture-based computer vision be used? Think about people who might not be able to interact with a computer in traditional ways.
👩💻 Exploration #3: Google Labs
Google Labs is a platform where Google shares and tests new experimental AI applications. These "experiments" demonstrate how AI can be applied to many different challenges and areas of interest! From generating new music to audio tours integrated with Google Maps, whatever you’re interested in, there’s probably a creative way to apply AI and machine learning!
Explore: Look around at the different experiments Google is working on. Find one that is particularly interesting to you and be prepared to share with the larger group!
Some of the projects showcased on Google Labs have become fully-fledged Google products or influenced other services. It's a place where Google continues to explore new concepts, gather user feedback, and iterate before committing to large-scale rollouts.
Imagine how you might combine AI with one of your passions. How could you use this technology to expand accessibility, spark creativity, or tackle real-world challenges?
Practice | The World of AI
Find an experiment on Google Labs to share with other scholars!
- Why did you choose this experiment to share?
- What underlying AI technology do you think this experiment uses?
- How can others play around with this experiment?
For a summary of this lesson, check out the 13. The World of AI One-Pager!