Launching soon — get early access:
AI Glossary

The Complete AI Dictionary

Master AI terminology from A to Z. 99 essential terms explained clearly, with real-world examples and connections to AI tools.

99 terms
A
AGENTS

AI Agent

An autonomous AI system that can perceive its environment, make decisions, and take actions to achieve specific goals.

70
A
ETHICS

AI Alignment

The challenge of ensuring AI systems behave in accordance with human intentions, values, and goals.

50
A
ETHICS

AI Safety

The field focused on preventing AI systems from causing unintended harm and ensuring they remain beneficial.

30
A
GENERAL

AI Wrapper

A product built on top of an existing foundation model through API calls, adding value at the application layer rather than training its own model.

30
A
AGENTS

Agent Memory

The mechanisms an AI agent uses to retain information within a session or across sessions, enabling context and personalization over time.

30
A
AGENTS

Agentic AI

AI systems that pursue goals autonomously — perceiving, planning, and using tools to act — rather than responding to one instruction at a time.

40
A
TECHNICAL

Application Programming Interface(API)

A set of protocols and tools that allows different software applications to communicate and share data.

30
A
GENERAL

Artificial Intelligence(AI)

The simulation of human intelligence processes by computer systems, including learning, reasoning, and self-correction.

40
A
TECHNICAL

Attention Mechanism

A neural network component that allows models to focus on relevant parts of the input when producing outputs.

70
A
AGENTS

Autonomous AI

AI systems capable of operating independently, making decisions and taking actions with minimal human oversight.

70
B
TECHNICAL

Batch Normalization(BatchNorm)

A technique that normalizes the inputs of each layer to have zero mean and unit variance, stabilizing and accelerating neural network training.

50
B
GENERAL

Benchmark

Standardized tests and datasets used to evaluate and compare AI model performance across different tasks.

50
B
ETHICS

Bias in AI

Systematic and unfair discrimination in AI outputs resulting from biased training data or algorithms.

50
B
MODELS

Bidirectional Encoder Representations from Transformers(BERT)

A transformer-based language model that reads text bidirectionally to better understand context and word relationships.

30
C
TECHNICAL

Chain-of-Thought(CoT)

A prompting technique that encourages AI models to break down complex problems into step-by-step reasoning.

30
C
TOOLS

Chatbot

A software application designed to simulate conversation with human users through text or voice interfaces.

40
C
TECHNICAL

Chunking

The process of splitting large documents into smaller, manageable pieces for embedding and retrieval in RAG systems.

30
C
GENERAL

Classification

A supervised learning task where the model assigns input data to predefined categories or classes.

60
C
GENERAL

Clustering

An unsupervised learning technique that groups similar data points together without predefined categories.

50
C
TECHNICAL

Computer Vision(CV)

A field of AI that enables computers to interpret and understand visual information from images and videos.

40
C
ETHICS

Constitutional AI(CAI)

An Anthropic-developed approach to training AI systems to be helpful, harmless, and honest using a set of principles rather than extensive human feedback.

50
C
LLM

Context Length

The maximum number of tokens a language model can process in a single prompt and response combined.

40
C
LLM

Context Window

The maximum amount of text (measured in tokens) that a language model can process in a single interaction.

40
C
TECHNICAL

Cosine Similarity

A metric that measures the similarity between two vectors by calculating the cosine of the angle between them, commonly used to compare text embeddings.

40
D
TECHNICAL

Deep Learning(DL)

A subset of machine learning that uses multi-layered neural networks to learn complex patterns from large amounts of data.

60
D
TECHNICAL

Diffusion Model

A type of generative model that creates data by learning to reverse a gradual noising process.

30
D
AGENTS

Digital Employee

An AI agent assigned to own a defined role or business outcome, performing multi-step work across systems with minimal human instruction.

30
D
TECHNICAL

Direct Preference Optimization(DPO)

A simpler alternative to RLHF that directly optimizes language models on human preference data without training a separate reward model.

30
D
TECHNICAL

Dropout

A regularization technique that randomly deactivates a percentage of neurons during training to prevent overfitting and improve generalization.

40
E
ETHICS

EU AI Act

The European Union's regulation that classifies AI systems by risk level and imposes obligations scaling with the potential for harm.

30
E
TECHNICAL

Embeddings

Dense vector representations of data (text, images, etc.) that capture semantic meaning in a numerical format.

40
E
ETHICS

Explainability(XAI)

The ability to understand and interpret how an AI model makes its predictions or decisions.

30
F
TECHNICAL

Few-Shot Learning

The ability of AI models to learn new tasks from just a few examples provided in the prompt.

30
F
TECHNICAL

Fine-Tuning

The process of further training a pre-trained model on a specific dataset to adapt it for a particular task or domain.

30
F
MODELS

Foundation Model(FM)

A large AI model trained on broad data that can be adapted to a wide range of downstream tasks.

30
F
TECHNICAL

Function Calling

An LLM capability where the model outputs structured JSON matching a predefined schema so an application can invoke external functions or APIs.

40
G
TECHNICAL

GPT-Generated Unified Format(GGUF)

A file format for storing quantized large language models optimized for efficient local inference on consumer hardware.

40
G
ETHICS

General Data Protection Regulation(GDPR)

The European Union's data-protection law, effective 2018, governing how organizations collect, process, and store EU residents' personal data.

30
G
GENERAL

Generative AI(GenAI)

AI systems that can create new content such as text, images, music, code, or video based on learned patterns.

60
G
MODELS

Generative Pre-trained Transformer(GPT)

A family of large language models developed by OpenAI that generate human-like text based on input prompts.

40
G
ETHICS

Guardrails

Safety mechanisms and constraints implemented to prevent AI systems from generating harmful, inappropriate, or off-topic content.

60
H
GENERAL

Hallucination

When an AI model generates plausible-sounding but factually incorrect or fabricated information.

40
H
ETHICS

Health Insurance Portability and Accountability Act(HIPAA)

A US law that sets national standards for protecting sensitive patient health information held by healthcare providers and their vendors.

30
H
AGENTS

Human-in-the-Loop(HITL)

A design pattern where a person reviews, approves, or corrects an AI system's outputs or actions before they take effect.

30
I
ETHICS

ISO 27001

An international standard specifying requirements for an information security management system (ISMS) to keep sensitive data secure.

30
I
TOOLS

Image Generation

AI technology that creates new images from text descriptions, sketches, or other inputs.

40
I
TECHNICAL

Inference

The process of using a trained AI model to make predictions or generate outputs from new input data.

30
J
ETHICS

Jailbreaking

Crafting inputs that trick an AI model into bypassing its safety guardrails and producing restricted or prohibited output.

30
L
TOOLS

LangChain

An open-source framework for building applications powered by large language models, providing tools for chains, agents, and memory.

50
L
LLM

Large Language Model(LLM)

A type of AI model trained on massive amounts of text data to understand, generate, and manipulate human language.

40
L
TOOLS

LlamaIndex

A data framework for connecting custom data sources to large language models, specializing in indexing and retrieval for RAG applications.

70
L
TECHNICAL

Low-Rank Adaptation(LoRA)

A parameter-efficient fine-tuning technique that trains small adapter layers instead of modifying all model weights.

50
M
TECHNICAL

MCP Server

A program that exposes data, tools, or prompts to AI applications over the Model Context Protocol, acting as the provider side of MCP.

50
M
GENERAL

Machine Learning(ML)

A subset of AI that enables systems to automatically learn and improve from experience without being explicitly programmed.

60
M
TECHNICAL

Mixture of Experts(MoE)

A neural-network design that routes each token to a few specialized sub-networks (experts), growing model capacity without growing per-token compute.

40
M
MODELS

Model Card

A short structured document describing an AI model's intended use, training data, performance, limitations, and ethical considerations.

30
M
TECHNICAL

Model Context Protocol(MCP)

An open standard, introduced by Anthropic, that lets AI applications connect to external data sources and tools through a uniform interface.

30
M
AGENTS

Multi-Agent System

An architecture in which multiple specialized AI agents coordinate, each handling part of a task, to achieve a larger goal together.

30
M
GENERAL

Multimodal AI

AI systems that can process and generate multiple types of data, such as text, images, audio, and video.

30
N
TECHNICAL

Named Entity Recognition(NER)

NLP task that identifies and classifies named entities (people, organizations, locations, etc.) in text.

30
N
TECHNICAL

Natural Language Processing(NLP)

A field of AI focused on enabling computers to understand, interpret, and generate human language.

30
N
TECHNICAL

Neural Network(NN)

A computing system inspired by biological neural networks, consisting of interconnected nodes that process information.

30
O
TOOLS

Ollama

An open-source tool that makes it easy to run large language models locally on your computer with a simple command-line interface.

30
O
GENERAL

Open Source AI

AI models and tools whose source code and/or weights are freely available for use, modification, and distribution.

40
O
MODELS

Open Weights

AI models whose trained parameters are publicly downloadable, so anyone can self-host and fine-tune them even without the training data or code.

30
O
TECHNICAL

Overfitting

When a machine learning model learns the training data too well, including noise and outliers, causing poor performance on new data.

40
P
ETHICS

PII Redaction

The automated detection and removal or masking of personally identifiable information before data is stored, logged, or sent to an AI model.

30
P
TECHNICAL

Parameters

The learned weights and biases in a neural network that determine how it processes input and produces output.

50
P
TOOLS

Prompt Engineering

The practice of designing and optimizing input prompts to get desired outputs from AI language models.

40
P
ETHICS

Prompt Injection

A security attack where malicious input tricks an AI into ignoring its original instructions and following attacker-controlled commands.

40
Q
TECHNICAL

Quantization

A technique that reduces model size and increases inference speed by using lower-precision number representations.

40
R
GENERAL

Regression

A supervised learning task where the model predicts a continuous numerical value rather than a category.

40
R
TECHNICAL

Regularization

Techniques that prevent overfitting by adding constraints or penalties to the learning process, encouraging simpler models.

60
R
TECHNICAL

Reinforcement Learning(RL)

A type of machine learning where an agent learns to make decisions by receiving rewards or penalties for its actions.

50
R
TECHNICAL

Reinforcement Learning from Human Feedback(RLHF)

A training technique that uses human preferences to fine-tune AI models to be more helpful, harmless, and honest.

60
R
TECHNICAL

Retrieval-Augmented Generation(RAG)

A technique that enhances LLM outputs by retrieving relevant information from external knowledge bases before generating responses.

30
S
ETHICS

SOC 2

A voluntary security-audit framework from the AICPA that reports on how a service organization manages customer data across five trust criteria.

30
S
TECHNICAL

Self-Hosted / On-Premise

A deployment model where software or an AI model runs on infrastructure the user controls rather than on the vendor's managed cloud.

30
S
TECHNICAL

Semantic Search

Search technology that understands the meaning and context of queries rather than just matching keywords.

40
S
TECHNICAL

Sentiment Analysis

NLP technique that identifies and extracts subjective information, determining whether text expresses positive, negative, or neutral sentiment.

40
S
TECHNICAL

Single Sign-On(SSO)

An authentication method that lets users log in once with one set of credentials to access multiple independent applications.

30
S
TECHNICAL

Software Development Kit(SDK)

A bundled set of libraries, tools, and documentation that helps developers build applications for a particular platform or service.

40
S
TOOLS

Speech-to-Text(STT)

AI technology that converts spoken audio into written text, also known as automatic speech recognition.

40
S
GENERAL

Supervised Learning

A machine learning approach where models learn from labeled training data to make predictions on new, unseen data.

40
S
TECHNICAL

Synthetic Data

Artificially generated data that mimics real-world data, used for training or testing AI models.

60
S
LLM

System Prompt

Hidden instructions given to an AI model that define its persona, behavior, capabilities, and constraints for a conversation.

50
T
TECHNICAL

Temperature

A parameter that controls the randomness and creativity of AI model outputs during text generation.

30
T
TOOLS

Text-to-Speech(TTS)

AI technology that converts written text into natural-sounding spoken audio.

30
T
TECHNICAL

Tokenization

The process of breaking down text into smaller units called tokens that AI models can process.

40
T
AGENTS

Tool Use

The capability of AI models to interact with external tools, APIs, and systems to accomplish tasks.

40
T
TECHNICAL

Top-P Sampling

A text generation technique that samples from the smallest set of tokens whose cumulative probability exceeds threshold P.

40
T
TECHNICAL

Transfer Learning

A technique where a model trained on one task is repurposed as the starting point for a model on a different but related task.

30
T
TECHNICAL

Transformer

A neural network architecture that uses self-attention mechanisms to process sequential data, revolutionizing NLP.

60
U
TECHNICAL

Underfitting

When a machine learning model is too simple to capture the underlying patterns in the data, resulting in poor performance on both training and test data.

30
U
GENERAL

Unsupervised Learning

A machine learning approach where models discover patterns and structures in unlabeled data without predefined categories.

60
V
TECHNICAL

Vector Database

A specialized database designed to store and efficiently query high-dimensional vector embeddings.

40
W
ETHICS

Web Content Accessibility Guidelines(WCAG)

A set of internationally recognized guidelines from the W3C for making web content accessible to people with disabilities.

30
W
TECHNICAL

Webhook

An automated HTTP callback that sends real-time data to a URL when an event occurs, letting systems notify each other without polling.

30
Z
TECHNICAL

Zero-Shot Learning

The ability of AI models to perform tasks they were not explicitly trained on, without any examples.

30
Stay Updated

AI Moves Fast. Stay Ahead.

New terms added weekly. Subscribe to get updates on the latest AI terminology and discover the AI tools shaping the future.