Master the language of Artificial Intelligence with clear explanations and real-world examples
Computers that can "think" like humans—learn, reason, and make decisions on their own.
Simulation of human intelligence in machines that can perform tasks like learning and problem-solving.
Siri or Alexa answering your voice commands.
A type of AI where computers learn from examples instead of being programmed for everything.
A subset of AI that enables systems to learn patterns from data and improve from experience.
Netflix recommending shows based on what you've watched.
Information like text, numbers, images, or audio that AI uses to learn and make decisions.
Raw or structured information that is used to train or evaluate AI systems.
A folder full of labeled photos of cats and dogs.
The final "brain" the AI creates after it finishes learning from data.
A trained algorithm that can make decisions or predictions based on input data.
An AI that can recognize hand-written digits after training.
Examples or past information we give to AI so it can learn.
A dataset of input-output pairs used to help a machine learning model learn patterns.
Teaching AI what an apple looks like by showing many apple pictures labeled as "apple."
The AI's guess or answer based on what it has learned.
The output of a model when it processes new data.
AI saying "this is a cat" when shown a new picture.
A list of instructions or a recipe that the computer follows to learn or solve problems.
A finite sequence of operations used for calculations, learning, and processing.
A sorting algorithm arranging names in order.
The instruction or question you give to an AI to get a response.
A text input provided to a language model to generate a specific type of response.
Typing "Write a poem about friendship" in ChatGPT.
AI that understands human language like speaking, writing, and reading.
A field of AI focused on enabling computers to understand, interpret, and generate human language.
Google Translate or auto-correct on your phone.
Splitting up sentences into smaller parts like words so AI can understand them.
The process of breaking down text into individual elements such as words or characters.
"Let's eat pizza" becomes ["Let's", "eat", "pizza"].
A way of designing AI to work like a human brain, using layers of small decision-makers (neurons).
A network of algorithms modeled after the human brain, consisting of layers of interconnected nodes.
AI that can recognize handwritten numbers on a check.
A special kind of learning using many layers of neural networks to understand complex things.
A subset of machine learning using multi-layered neural networks to process large and complex datasets.
YouTube recommending videos based on what you watch.
A really big AI trained on a lot of text so it can read, write, and respond like a human.
A deep learning model trained on massive text datasets to perform language-related tasks.
GPT-4 powering ChatGPT or Google Gemini.
AI that can create new content like stories, images, or songs.
AI systems that generate new data based on patterns learned during training.
DALL·E creating a drawing of a cat flying a plane.
A virtual assistant or talking robot that answers questions and helps users.
An application that simulates conversation using NLP to interact with users.
A shopping website's live support bot.
A big collection of information used for training or testing AI.
An organized set of data used in machine learning for training and evaluation.
A spreadsheet with 1,000 labeled animal photos.
AI that decides which category something belongs to.
A machine learning task where the output is a discrete class label.
AI classifying emails as "spam" or "not spam."
Teaching AI by showing it examples with the correct answer.
A machine learning method using labeled data to train the model.
Training AI to recognize dogs by labeling images as "dog."
AI learns patterns without being told the right answers.
A machine learning method where the model is given data without labeled outcomes.
AI grouping customers by similar shopping habits.
AI groups similar things together on its own.
An unsupervised learning technique that organizes data into similar groups.
Grouping songs by mood or genre without labels.
AI learns by trying things out and getting rewards or punishments—like training a pet.
A learning method where an agent interacts with an environment and learns by receiving feedback (reward or penalty).
AI playing chess and learning to win by trial and error.
A detail or fact about something that helps AI make decisions.
An individual measurable property or input variable used in modeling.
House price prediction uses features like size and location.
The correct answer AI uses while learning from examples.
The outcome or category assigned to training data in supervised learning.
Image labeled "dog" so AI knows it's a dog.
The process of teaching AI using data so it can learn patterns.
Feeding data into a model and adjusting parameters to learn patterns.
Teaching AI what fruits look like by showing many labeled fruit images.
When AI gives an answer after training—like solving a new problem.
Using a trained model to make predictions or decisions on new data.
AI recognizing a cat in a brand-new photo.
Turning words or images into numbers so AI can understand them.
A way of converting data into numerical vectors representing similarity or meaning.
"King" and "Queen" are close together in embedding space.
The settings inside an AI model that get adjusted during learning.
Internal values (like weights) that a model learns during training.
GPT-4 has billions of parameters controlling how it responds.
Teaching an already trained AI to do a more specific job.
Adjusting a pre-trained model using task-specific data.
Tuning a chatbot to speak like a lawyer or doctor.
A way for apps to talk to each other or to an AI model.
A set of tools and protocols that let different software systems communicate.
Using ChatGPT API to add AI to your website or app.
When AI makes unfair or unbalanced decisions due to bad or one-sided data.
Systematic error introduced by prejudice in training data or assumptions.
AI that prefers certain names when screening resumes.
How often the AI gets things right.
The ratio of correct predictions to total predictions made.
AI correctly identifies 95 out of 100 animals.
How many of AI's "positive" answers were actually correct.
True positives divided by the number of total predicted positives.
AI marks 10 emails as spam, 8 actually were → 80% precision.
How many correct answers the AI could find out of all the right ones.
True positives divided by the number of total actual positives.
AI finds 7 out of 10 spam emails → 70% recall.
A balance between precision and recall—a way to measure overall performance.
The harmonic mean of precision and recall.
Helps measure how well AI performs on both accuracy and coverage.
When AI learns too much from examples and struggles with new situations.
A modeling error where the AI performs well on training data but poorly on new data.
Student memorizes old test answers but fails new ones.
When AI doesn't learn enough from the data.
A modeling error where the model is too simple to learn the pattern in data.
Student who barely studies and gets many answers wrong.
When AI confidently gives an answer that's totally wrong or made up.
When an AI model generates false or misleading outputs that don't match reality.
ChatGPT says "Apple is the capital of France."
The company that created ChatGPT and other popular AI tools.
An AI research company focused on ensuring safe and beneficial AI development.
The makers of GPT-4, DALL·E, Whisper, etc.
Making sure AI is used fairly, safely, and responsibly.
The field that ensures AI systems are built and used in ways that are morally and socially acceptable.
Ensuring AI doesn't discriminate or spread fake news.
AI that can explain how and why it made a decision.
Methods and tools that help interpret AI decisions and outputs.
AI explaining why it rejected someone's loan application.