Reference

AI Terminology Glossary

Every AI term explained in plain English. Click any term to read the full definition, see real-world examples, and practice unscrambling it.

A B C D E F G H I J L M N P Q R S T V W

A

An AI agent is a system where a language model is given tools, memory, and the abilit...

Agentic refers to AI systems that operate autonomously over extended time horizons, m...

Artificial General Intelligence (AGI) refers to a hypothetical AI system that can per...

AI alignment is the challenge of ensuring that AI systems pursue goals and behave in ...

An API (Application Programming Interface) is the connection layer that allows develo...

Attention is a mechanism in the Transformer architecture that allows a model to focus...

B

A benchmark is a standardised test or dataset used to measure and compare AI model pe...

AI bias refers to systematic errors or unfair patterns in AI outputs that reflect pre...

C

Chain of Thought

Chain-of-thought (CoT) prompting is a technique where an AI model is instructed to re...

Compute refers to the processing power used to train or run AI models, typically meas...

Constitutional AI

Constitutional AI is a training technique developed by Anthropic in which an AI model...

The context window is the maximum amount of text — measured in tokens — that an AI mo...

D

Diffusion Model

A diffusion model is a type of generative AI model that learns to create new images, ...

Knowledge distillation is a model compression technique where a smaller "student" mod...

E

An embedding is a numerical representation of text, images, or other data as a vector...

Emergent capabilities are abilities that appear in large AI models but are not presen...

F

Fine-tuning is the process of taking a pre-trained AI model and continuing to train i...

A frontier model refers to the most capable AI models at the current state of the art...

G

GPT stands for Generative Pre-trained Transformer. It is a family of large language m...

A GPU (Graphics Processing Unit) is a specialised chip originally designed for render...

Grounding in AI refers to connecting a model's outputs to verified, real-world facts,...

Guardrails are rules, filters, classifiers, or constraints built into an AI system to...

H

AI hallucination refers to when an AI model generates information that is factually i...

I

Inference is the process of running a trained AI model to generate predictions or out...

J

A jailbreak is a prompt, technique, or sequence of inputs designed to bypass an AI mo...

L

Latency in AI systems refers to the time delay between sending a request to a model a...

A Large Language Model (LLM) is an artificial intelligence system trained on massive ...

LoRA (Low-Rank Adaptation) is a parameter-efficient fine-tuning technique that enable...

M

A multimodal AI model is one that can process and generate multiple types of data — s...

N

Natural Language Processing (NLP) is the field of artificial intelligence focused on ...

P

Perplexity is a measurement of how well a language model predicts a text sample. A lo...

A prompt is the input text that a user or developer provides to an AI model to instru...

Prompt Engineering

Prompt engineering is the practice of designing and refining the text inputs given to...

Q

Quantization is a technique for reducing the size and memory requirements of AI model...

R

Retrieval-Augmented Generation (RAG) is a technique that combines a search or retriev...

Retrieval in the context of AI refers to the process of fetching relevant documents, ...

Reinforcement Learning from Human Feedback (RLHF) is a training technique where human...

S

Scaling refers to the empirical observation that AI model performance improves predic...

Synthetic data is data generated by AI models rather than collected from real-world s...

T

Temperature is a parameter that controls the randomness of an AI model's outputs. A t...

A token is the basic unit of text that a language model processes. Tokens are roughly...

Training is the process by which an AI model learns patterns from data by repeatedly ...

The Transformer is a neural network architecture introduced in the 2017 paper "Attent...

V

Vector Database

A vector database is a type of database specifically designed to store and search emb...

W

An AI workflow is an automated sequence of tasks that chains together multiple AI mod...

Learn these terms with daily challenges

AI Terminology Scrambler uses spaced repetition and daily word scrambles to help you build real AI vocabulary — not just memorise definitions.

Play the daily challenge →