AI Glossary
Key Terms & Concepts in Artificial Intelligence
Master the language of AI with simple, expert explanations of essential artificial intelligence terms, models, and tools.
A
A future AI that can think and learn like a human in any situation.
A future AI that can think and learn like a human in any situation.
A future AI that can think and learn like a human in any situation.
Smart AI systems that plan and act on their own to get things done.
Smart AI systems that plan and act on their own to get things done.
Smart AI systems that plan and act on their own to get things done.
Making sure AI follows human goals, values, and safety.
Making sure AI follows human goals, values, and safety.
Making sure AI follows human goals, values, and safety.
Making sure AI is used in fair, safe, and responsible ways.
Making sure AI is used in fair, safe, and responsible ways.
Making sure AI is used in fair, safe, and responsible ways.
Testing how well an AI system works, behaves, and meets its goals.
Testing how well an AI system works, behaves, and meets its goals.
Testing how well an AI system works, behaves, and meets its goals.
Finding when AI makes things up so the output can be corrected.
Finding when AI makes things up so the output can be corrected.
Finding when AI makes things up so the output can be corrected.
The part of AI that remembers things to act smarter over time.
The part of AI that remembers things to act smarter over time.
The part of AI that remembers things to act smarter over time.
Making sure AI doesn’t cause harm and stays under human control.
Making sure AI doesn’t cause harm and stays under human control.
Making sure AI doesn’t cause harm and stays under human control.
AI is the ability of machines to mimic human intelligence.
AI is the ability of machines to mimic human intelligence.
AI is the ability of machines to mimic human intelligence.
AI systems that act and make decisions by themselves to complete tasks.
AI systems that act and make decisions by themselves to complete tasks.
AI systems that act and make decisions by themselves to complete tasks.
B
When AI makes unfair decisions because of bad or unbalanced data.
When AI makes unfair decisions because of bad or unbalanced data.
When AI makes unfair decisions because of bad or unbalanced data.
C
AI that helps machines see and understand pictures or video.
AI that helps machines see and understand pictures or video.
AI that helps machines see and understand pictures or video.
AI trained to follow written rules, not just human feedback.
AI trained to follow written rules, not just human feedback.
AI trained to follow written rules, not just human feedback.
The maximum number of tokens an AI model can process at once.
The maximum number of tokens an AI model can process at once.
The maximum number of tokens an AI model can process at once.
AI solves problems step by step by showing its thinking clearly.
AI solves problems step by step by showing its thinking clearly.
AI solves problems step by step by showing its thinking clearly.
D
Tagging data so AI knows what it is and can learn from it.
Tagging data so AI knows what it is and can learn from it.
Tagging data so AI knows what it is and can learn from it.
Advanced neural networks that analyze complex data like images or speech.
Advanced neural networks that analyze complex data like images or speech.
Advanced neural networks that analyze complex data like images or speech.
E
Translating content into numbers AI can work with.
Translating content into numbers AI can work with.
Translating content into numbers AI can work with.
Turning words or data into numbers so AI can understand meaning.
Turning words or data into numbers so AI can understand meaning.
Turning words or data into numbers so AI can understand meaning.
Helping people understand how AI made a choice or prediction.
Helping people understand how AI made a choice or prediction.
Helping people understand how AI made a choice or prediction.
F
AI that learns a task from just a few examples.
AI that learns a task from just a few examples.
AI that learns a task from just a few examples.
Training a pre-trained model to make it better at one task.
Training a pre-trained model to make it better at one task.
Training a pre-trained model to make it better at one task.
G
AI that creates new text, images, music, or code from data.
AI that creates new text, images, music, or code from data.
AI that creates new text, images, music, or code from data.
A way AI models learn by fixing mistakes step by step.
A way AI models learn by fixing mistakes step by step.
A way AI models learn by fixing mistakes step by step.
h
When AI makes up answers that sound real but aren’t true.
When AI makes up answers that sound real but aren’t true.
When AI makes up answers that sound real but aren’t true.
I
The moment an AI responds to your input—its real-time thinking.
The moment an AI responds to your input—its real-time thinking.
The moment an AI responds to your input—its real-time thinking.
The size of your prompt measured in tokens—not characters or words.
The size of your prompt measured in tokens—not characters or words.
The size of your prompt measured in tokens—not characters or words.
J
K
Teaching a small AI model to act like a big one.
Teaching a small AI model to act like a big one.
Teaching a small AI model to act like a big one.
L
AI trained on massive datasets to understand and generate text.
AI trained on massive datasets to understand and generate text.
AI trained on massive datasets to understand and generate text.
A faster, cheaper way to fine-tune large AI models.
A faster, cheaper way to fine-tune large AI models.
A faster, cheaper way to fine-tune large AI models.
M
Teaches systems to learn and improve from data without being explicitly programmed.
Teaches systems to learn and improve from data without being explicitly programmed.
Teaches systems to learn and improve from data without being explicitly programmed.
Public reports that explain how AI models were built and what they do.
Public reports that explain how AI models were built and what they do.
Public reports that explain how AI models were built and what they do.
Making AI models smaller and faster without losing too much quality.
Making AI models smaller and faster without losing too much quality.
Making AI models smaller and faster without losing too much quality.
AI that works with text, images, audio, and more at the same time.
AI that works with text, images, audio, and more at the same time.
AI that works with text, images, audio, and more at the same time.
N
AI model inspired by the brain, used to detect patterns and make decisions.
AI model inspired by the brain, used to detect patterns and make decisions.
AI model inspired by the brain, used to detect patterns and make decisions.
Allows machines to understand and respond to human language.
Allows machines to understand and respond to human language.
Allows machines to understand and respond to human language.
O
AI models with public weights that anyone can use, study, or customize.
AI models with public weights that anyone can use, study, or customize.
AI models with public weights that anyone can use, study, or customize.
P
AI that finds shapes, trends, or repeated patterns in data.
AI that finds shapes, trends, or repeated patterns in data.
AI that finds shapes, trends, or repeated patterns in data.
Initial training that teaches AI general knowledge from big data.
Initial training that teaches AI general knowledge from big data.
Initial training that teaches AI general knowledge from big data.
A method where AI follows a series of prompts to reach better answers.
A method where AI follows a series of prompts to reach better answers.
A method where AI follows a series of prompts to reach better answers.
Crafting inputs to guide AI models like ChatGPT more effectively.
Crafting inputs to guide AI models like ChatGPT more effectively.
Crafting inputs to guide AI models like ChatGPT more effectively.
Tricking an AI into misbehaving by altering or hijacking its prompt.
Tricking an AI into misbehaving by altering or hijacking its prompt.
Tricking an AI into misbehaving by altering or hijacking its prompt.
Q
R
An AI method that combines search with text generation for better answers.
An AI method that combines search with text generation for better answers.
An AI method that combines search with text generation for better answers.
AI framework for thinking step by step and then taking action.
AI framework for thinking step by step and then taking action.
AI framework for thinking step by step and then taking action.
Testing AI by trying to break it and find harmful or risky behavior.
Testing AI by trying to break it and find harmful or risky behavior.
Testing AI by trying to break it and find harmful or risky behavior.
AI learns by trying actions and getting rewards or penalties.
AI learns by trying actions and getting rewards or penalties.
AI learns by trying actions and getting rewards or penalties.
Training AI using human ratings to improve its answers and behavior.
Training AI using human ratings to improve its answers and behavior.
Training AI using human ratings to improve its answers and behavior.
Building machines that can move, sense, and think using AI.
Building machines that can move, sense, and think using AI.
Building machines that can move, sense, and think using AI.
S
AI that learns from data without needing human-labeled answers.
AI that learns from data without needing human-labeled answers.
AI that learns from data without needing human-labeled answers.
AI search that looks for meaning, not just matching words.
AI search that looks for meaning, not just matching words.
AI search that looks for meaning, not just matching words.
A virtual space where AI can practice or be tested safely.
A virtual space where AI can practice or be tested safely.
A virtual space where AI can practice or be tested safely.
T
Breaking text into small pieces so AI can read and understand it.
Breaking text into small pieces so AI can read and understand it.
Breaking text into small pieces so AI can read and understand it.
AI connected to tools or apps so it can take actions or find answers.
AI connected to tools or apps so it can take actions or find answers.
AI connected to tools or apps so it can take actions or find answers.
The data an AI model learns from during training.
The data an AI model learns from during training.
The data an AI model learns from during training.
AI reuses what it learned from one task to do another.
AI reuses what it learned from one task to do another.
AI reuses what it learned from one task to do another.
Making it clear how AI works, what it uses, and who controls it.
Making it clear how AI works, what it uses, and who controls it.
Making it clear how AI works, what it uses, and who controls it.
U
AI finds patterns in data without needing labeled answers.
AI finds patterns in data without needing labeled answers.
AI finds patterns in data without needing labeled answers.
V
A special database that finds similar things using meaning, not words.
A special database that finds similar things using meaning, not words.
A special database that finds similar things using meaning, not words.
W
Y
Start Building Your Business Today
Learn how to create, automate, and grow using the most powerful technology of our time.
Start Building Your Business Today
Learn how to create, automate, and grow using the most powerful technology of our time.
Start Building Your Business Today
Learn how to create, automate, and grow using the most powerful technology of our time.