Welcome! Let’s dive into the world of AI Lab, a cool tool that helps you use data to train a machine learning model. Once your model is ready, you can bring it into App Lab and use it in your projects. Let’s explore how it works!
The first step in AI Lab is picking a dataset that interests you. Imagine you’re curious about animals, so you choose the zoo dataset. This dataset has lots of information about different animals and their features. Your task is to decide which features might help you make interesting predictions.
Let’s say you want to predict if an animal is a predator. You start by selecting features that might help with this prediction. For example, you might think about whether having feathers is related to being a predator. However, you find that feathers don’t really help, so you skip that feature.
Instead, you notice that being venomous is often linked to being a predator, so you choose that feature. You also decide to use the animal’s class, like whether it’s a mammal, fish, or bird, to help with your prediction. With these features selected, you’re ready to train your model.
Now, let’s meet AI Bot, a smart tool that helps train your model. AI Bot takes your dataset and splits it into two parts: training data and testing data. The training data is used to find patterns and make predictions, while the testing data checks how accurate those predictions are.
As AI Bot works with the training data, it learns to recognize patterns that help it decide if an animal is a predator. Once the model is trained, AI Bot tests it with the testing data to see how well it performs. In our example, AI Bot achieved 63% accuracy, while our actual model did even better with 72% accuracy!
After training your model, it’s important to create a model card. Think of a model card like a nutrition label on food. It provides details about the features used, potential applications, and any limitations of your model. This helps others understand how your model works and any warnings they should know about.
Once your model card is ready, you can import your model into App Lab. In the manage AI models section, you’ll find your model, “Predator,” ready to be imported. When you bring it into App Lab, some design elements and programming code are already set up for you.
Now, you can run your program, input data, and see if your model predicts whether an animal is a predator. It’s a fun way to see machine learning in action!
We’re thrilled to offer new resources for learning about artificial intelligence and machine learning in classrooms. To explore more, visit code.org. By clicking the “Explore the Module” button, you’ll find a module designed for students like you. AI Lab is just one part of a larger curriculum that supports learning about machine learning.
Check out the calendar and click on any lessons to access additional resources and lesson plans. Happy learning!
Choose a dataset that interests you, such as the zoo dataset mentioned in the article. Identify different features of the animals and discuss with your classmates which features might help in making predictions. This will help you understand the importance of selecting relevant data for machine learning.
Play a game where you guess which features might be useful for predicting if an animal is a predator. For example, consider features like being venomous or the animal’s class. Discuss why some features are more relevant than others and how they can affect the accuracy of predictions.
Use AI Lab to train a machine learning model with your chosen features. Work in pairs to split your dataset into training and testing data. Observe how AI Bot learns from the training data and test its accuracy. Compare your results with your classmates to see who achieved the highest accuracy.
Design a model card for your trained model. Include details about the features you used, potential applications, and any limitations. Share your model card with the class and explain how it helps others understand your model’s capabilities and warnings.
Import your trained model into App Lab and use it in a simple project. Design a program where you input data about an animal and see if your model predicts it as a predator. Share your project with classmates and discuss the outcomes and any improvements you could make.
Sure! Here’s a sanitized version of the transcript:
—
Hi, my name is Erin Bond, and I’m one of the software engineers who helped build AI Lab. AI Lab is a tool that allows you to use data to train a machine learning model that can then be imported into App Lab and used in a project.
The first thing we do in AI Lab is decide which dataset we’re interested in. For example, I find the zoo dataset intriguing. I can see the columns of information included in that dataset and some examples of what’s in there. In this case, we have various animals and different features about them. It’s my job to decide which of these features might make an interesting or useful prediction.
I think I want to figure out whether an animal is a predator or not, so I’ll click continue. Now I need to choose which other features I think will be helpful in predicting whether an animal is a predator. For instance, I consider whether feathers might indicate if an animal is a predator. However, there’s not much correlation between feathers and predatory behavior, so I don’t think I’ll use that feature.
On the other hand, venom seems useful, as there is a high correlation between being venomous and being a predator. So, I’ll add that feature. I also want to use the animal’s class—whether it’s a mammal, fish, or bird—to help me predict if it’s a predator. Therefore, I will predict predator status based on venomous and class features, and now I’m going to train the model.
Hi, my name is Dan, and I’m a curriculum writer with Code.org. I helped write the unit on AI and machine learning that uses AI Lab and AI Bot to train models. The screens we see now are crucial parts of the process where AI Bot learns to make decisions based on data.
AI Bot takes our original dataset and splits it into two parts. A large portion is called the training data, which AI Bot studies to find patterns and make predictions. A smaller portion is called the testing data, which AI Bot will use later to evaluate its decision-making accuracy.
As AI Bot processes the training data, it looks for patterns to create a machine learning model. Each row of data influences how AI Bot will make future decisions about whether an animal is a predator. Once AI Bot has trained its model, it examines the testing dataset to see how well it performs. Since we already know the correct predictions for this testing data, we can compare AI Bot’s predictions to the actual outcomes.
In this example, AI Bot achieved 63% accuracy. Now, let’s see how our actual model performed. It achieved 72% accuracy. We can review the predictions AI Bot got correct and identify where it struggled, such as confusing certain animals.
Once we’ve prepared our model, we need to create something called a model card. A model card provides important context about the decisions made during training. It includes information about the features deemed important, potential uses for the model, and any limitations or warnings for real-world applications.
Model cards can be thought of like nutrition labels on food, providing insights into how the product was prepared and any relevant warnings. After saving the model card, we can import our model into App Lab.
In the manage AI models section, I can see my model, “Predator,” and import it. When I import the model, some design elements are pre-populated for a form that resembles the prediction screen in AI Lab, along with the necessary programming code.
When I run my program, I can input data, and it predicts whether the animal is a predator. We’re excited about the new resources we’ve created for teaching artificial intelligence and machine learning in classrooms. To learn more, you can visit code.org.
If you press the “Explore the Module” button, you’ll be directed to the new module designed for middle school students. AI Lab is just one part of a larger curriculum we’ve developed to support machine learning in classrooms. You can view the calendar and click on any lessons to access additional resources and lesson plans.
—
This version removes any informal language and personal identifiers while maintaining the core content and structure.
AI – AI stands for Artificial Intelligence, which is the ability of a computer or a robot to perform tasks that usually require human intelligence. – Example sentence: AI can help computers recognize faces in photos.
Lab – A lab is a place equipped for scientific experiments, research, or teaching, often used for testing new computer technologies. – Example sentence: Our computer lab has the latest software for learning about AI.
Model – In AI, a model is a program or algorithm that has been trained to recognize patterns or make decisions based on data. – Example sentence: The AI model can predict the weather by analyzing past data.
Dataset – A dataset is a collection of data that is used to train or test an AI model. – Example sentence: We used a large dataset of animal images to teach the AI to identify different species.
Predictions – Predictions are the outcomes or results that an AI model forecasts based on the data it has been given. – Example sentence: The AI made predictions about which movies I might like based on my viewing history.
Training – Training is the process of teaching an AI model by feeding it data so it can learn to make accurate predictions or decisions. – Example sentence: We spent weeks training the AI to understand different accents in speech.
Bot – A bot is a software application that runs automated tasks over the internet, often used to simulate conversation or perform repetitive tasks. – Example sentence: I chatted with a customer service bot to get help with my computer issue.
Features – Features are individual measurable properties or characteristics used by an AI model to make decisions. – Example sentence: The AI uses features like color and shape to identify objects in pictures.
Accuracy – Accuracy is a measure of how often an AI model makes correct predictions or decisions. – Example sentence: The AI’s accuracy improved after we added more data to its training set.
Code – Code is a set of instructions written in a programming language that tells a computer what to do. – Example sentence: I wrote code to create a simple game using Python.
