AI Basics
In AI vocabulary, this is the "trained brain" of an AI system that holds everything the AI learned and uses it to make predictions.
What is a model?
A computer sees images as a grid of these tiny units, each with a number from 0 (black) to 255 (white) representing brightness.
What are pixels?
Name one everyday app or device that uses AI to give you personalized recommendations or recognize your voice.
What is Spotify / YouTube / Siri / Netflix / Google Maps? (multiple answers are acceptable!)
This core principle of AI says that if you feed a model bad, wrong, or incomplete data, you will always get bad results; just like a recipe made with rotten ingredients.
What is "Garbage In, Garbage Out" (GIGO)?
This happens when an AI gives an answer that sounds confident and correct, but is actually wrong, often because its training data was incomplete or outdated.
What is an AI hallucination?
This type of AI learns from examples and gets smarter as it sees more data, instead of being programmed for every task.
What is Machine Learning (ML)?
A neural network is inspired by the human brain. It's made of these individual units that each take in numbers, do a small calculation, and pass the result to the next layer.
What are neurons (or nodes)?
Students used this AI music tool to generate original songs just by typing a text prompt, such as describing a style and/or mood. Then, it creates the audio for you.
What is Suno (an AI Music Generator)?
An AI model isn't just about having a lot of data; it needs a lot of this kind of data specifically, because quantity alone won't make a good model.
What is good data (or accurate / labeled / high-quality data)?
AI should be a ____, not a replacement for your own thinking, especially in school.
What is a tool?
This is the data used to teach an AI model, like practice problems before a math test.
What is training data?
This field of AI lets computers "see" and interpret the world. It powers self-driving cars, medical scans, and facial recognition.
What is Computer Vision (CV)?
This type of AI creates brand-new content, such as images, music, or text, from a description you type in, called a prompt.
What is Generative AI?
When building an AI model, your data gets split into three groups. Training data = teaches the model. ___ data = checks it while still learning. Test data = gives the final grade.
What is validation data?
These are realistic but fake videos or images where AI replaces someone's face or voice. It is a serious threat to media trust and online safety.
What are deepfakes?
This is new data an AI model has never seen before, used to check whether it learned correctly.
What is testing data?
This computer vision task identifies what is in an image and where it is located, often using bounding boxes. Steps are 1) Get data, 2) Annotate data, & 3) Train the model and test it after.
What is object detection?
AI can sound totally convincing, but this is the habit every smart AI user should have; cross-checking what AI tells you against trustworthy websites, books, or experts before believing it.
What is verifying your sources (or fact-checking)?
This type of AI model predicts a number like a puppy's age based on its weight, by drawing the best-fit line through data points.
What is Linear Regression (LinReg)?
When AI makes unfair decisions because of patterns in its training data, this problem is called this.
What is bias?
These are the rules or step-by-step instructions that tell a computer how to learn from data and solve a problem.
What is an algorithm (or program)?
This computer vision task identifies what is in an image, such as determining that a picture contains a cat.
What is classification?
Students used this platform/tool to train a custom object-detection model by uploading and labeling classroom photos (hint: bus).
What is Roboflow?
This AI model is made of of an input layer, hidden layer(s), and an output layer. They are connected via neurons that work together to learn patterns and make predictions.
What is a Neural Network (NN)?
This is the practice of writing clear, detailed instructions so an AI gives you better answers.
What is prompt engineering?