AI Mechanics
AI Prompting
AI Ethics and Risks
AI Bias and Fairness
P.A.U.S.E
100

AI doesn’t "think" like a human; instead, it looks for these repeating things in data to learn.

Patterns

100

This is the specific name for the command or question you type into an AI.

Prompt

100

This is a video or audio recording created by AI that makes a person appear to say or do something they never did.

Deepfake

100

When an AI gives an answer that is unfair to a certain group of people, it is called this.

Bias

100

The "P" stands for this—the most important input you give the AI.

Prompt

200

This type of AI structure is inspired by the human brain and uses layers to solve problems.

Neural Networks

200

To get a better answer, you should tell the AI to act as a specific character, like a "Historian" or "Coach." This is called giving the AI a...

Role

200

When an AI confidently tells you a fact that is completely made up and false.

Hallucination

200

AI bias usually happens because this thing (which the AI learns from) is incomplete or old.

Training Data

200

The "S" reminds you that you must never trust an AI fact without checking these.

Sources

300

When ChatGPT writes a sentence, it isn't telling the truth; it is just using math to do this with the next word.

Predictions

300

Instead of one AI doing everything, this is a group of specialized AIs working together to solve a big goal.

AI Agent Team

300

You should assume that everything you type into a public AI chatbot is doing this.

Saving your Data

300

Humans use this "c-word" and empathy to make decisions

Context
300

The "E" stands for this—your job to change, fix, and improve what the AI writes.

Edit

400

To an AI, the lines, shapes, and colors in a picture act as these "hints" to tell it what the image is.

Features
400

If you want the AI to sound "excited" or "serious" or "funny," you need to change this in your prompt.

Tone

400

Before using AI for any school assignment, the "A" in P.A.U.S.E. reminds you to do this.

Ask

400

Bias is often caused by the training data not being this, meaning it doesn't represent all people, places, or perspectives.

Representative or Diverse

400

The "U" reminds us to look out for "Unfairness," which is another word for this.

Bias

500

This "M.L." is the process where computers learn from data without being explicitly programmed for every single step.

Machine Learning

500

A "Manager Agent" breaks a big goal down into smaller tasks and gives them to these

Sub-Agents

500

Using AI to write your entire essay without doing any work yourself is considered this

Plagiarism

500

This is the only "person" who can fix AI bias by checking the answer and adding fairness.

Human

500

The "A" stands for this—because different teachers have different rules.

Ask