Artificial intelligence (AI)
A broad name for computer systems designed to perform tasks that usually require human judgment, such as recognizing patterns, generating language, or planning a route. “AI” describes a family of techniques, not a single mind.
Machine learning
A way of building software that learns patterns from examples instead of receiving every rule by hand. The examples and the design of the system shape what it can recognize—and what it misses.
Model
A trained mathematical system that turns inputs into outputs. A model can classify an image, predict the next word, recommend a video, or generate a response. It is not a database of guaranteed facts.
Language model
A model trained to work with language. A large language model (LLM) predicts likely sequences of tokens—small pieces of text—based on patterns in its training and the conversation context.
Prompt
The instructions, question, examples, and context a person gives a model. A prompt can improve an answer by stating the goal, audience, constraints, and what should happen when the model is uncertain.
Hallucination
An inaccurate or invented output presented in a plausible way. It can look like a citation, quote, number, or confident explanation. Treat surprising claims as a request to verify, not as a reason to trust the tone.
Bias
A systematic pattern that disadvantages or misrepresents people, places, or viewpoints. Bias can enter through training data, labels, design choices, or the context in which an output is used. Checking whose perspective is missing is a useful first step.
Training data
Examples used to adjust a model’s internal parameters. Training data can include text, images, audio, or other records, depending on the system. “The model saw a pattern” does not mean it knows the source, rights, or context of every example.
Context window
The amount of text or other input a model can consider in one interaction. When a conversation is long, earlier details may be left out or summarized, so important instructions should be restated.
Human-in-the-loop
A design in which a person reviews, guides, or approves an AI-assisted decision. For children, human involvement should increase when a topic is personal, high-stakes, or difficult to verify.
The useful mental model: AI is a powerful pattern tool. It can help you explore and practise, but a person remains responsible for the purpose, the evidence, and the consequences.