Welcome to the definitive guide to understanding the universe of Artificial Intelligence. This glossary is designed to take you from the most basic concepts to the most advanced ideas that are shaping our future.
Whether you are a beginner or an expert, this resource will serve as a guide to understand how AI impacts your daily life. Explore and discover how to apply these concepts in your projects! If you have suggestions, leave them in the comments.
Fundamentals of AI
Essential concepts that form the basis of Artificial Intelligence.
Artificial Intelligence (AI)
Field of computer science dedicated to creating systems capable of performing tasks that typically require human intelligence, such as learning, reasoning, problem solving, perception, and language understanding.
Algorithm
A finite sequence of well-defined instructions or rules that are followed to solve a problem or perform a task. In AI, algorithms are the “brain” that processes data and learns from it.
(AI) Model
It is the result of training an algorithm with a set of data. A model is a mathematical structure (such as a neural network) that has learned to recognize patterns or make predictions from data.
- Example: A model trained with thousands of cat photos can identify if a new photo contains a cat.
Artificial Narrow Intelligence (ANI – Artificial Narrow Intelligence)
Also known as weak AI, this is the type of AI we see today. It is designed and trained for a specific task and cannot operate outside its defined scope.
- Example: Siri, Netflix recommendation systems or facial recognition software.
Artificial General Intelligence (AGI – Artificial General Intelligence)
A type of theoretical AI that has the ability to understand, learn and apply its intelligence to solve any problem, similar to a human being. Its development has not yet been achieved.
Key Disciplines of AI
The main branches and subfields that make up the study of AI.
Machine Learning (Machine Learning)
Subfield of AI that focuses on developing algorithms that allow computers to learn from data and improve their performance with experience, without being explicitly programmed for each task.
Deep Learning (Deep Learning)
An advanced Machine Learning technique that uses multi-layered neural networks (deep neural networks) to analyze complex data such as images, sound, and text. It is the engine of the most recent advances in AI.
Artificial Neural Network
A computational model inspired by the structure and functioning of the human brain. It consists of interconnected nodes (neurons) organized in layers that process information and recognize patterns.
Natural Language Processing (NLP)
Branch of AI that allows machines to understand, interpret, generate and respond to human language (text and voice) in a useful and coherent way.
- Example: Automatic translators, chatbots and sentiment analysis on social networks.
Computer Vision (Computer Vision)
Field of AI that trains machines to interpret and understand the visual world. It allows computers to “view” and analyze the content of images and videos.
- Example: Face unlocking on phones, autonomous cars that identify obstacles or medical diagnosis from x-rays.
Multimodal AI
AI systems capable of processing and relating information from multiple types of data (or “modalities”) at the same time, such as text, images, audio and video.
- Example: Generate a text description of what happens in a video.
Data: The Fuel of AI
Terms related to the information that feeds AI models.
Data (Data)
The raw material of AI. It is the information (text, numbers, images, etc.) that is used to train, validate and test machine learning models.
Big Data
Extremely large and complex data sets that exceed the ability of traditional software to be captured, managed and processed in a reasonable time.
Dataset (Dataset)
A structured collection of data used for a specific purpose, usually divided into training, validation and test sets to develop an AI model.
Data Mining (Data Mining)
The process of discovering patterns, correlations, and anomalies within large data sets to predict outcomes.
Tokenization
The process of dividing text into smaller units called “tokens” (which can be words, parts of words, or characters). It is a fundamental step so that language models can process the text.
- Example: The phrase “The AI is here” is tokenized in [“The”, “IA”, “is”, “here”].
Embeddings (Insertion Vectors)
Numerical representations of data (such as words or images) in a multi-dimensional space. They allow models to capture semantic and contextual relationships between data.
- Example: The words “king” and “queen” will have embeddings close in this space, as will “dog” and “cat”.
Types of Learning and Models
The different ways models learn and the most common architectures.
Supervised Learning
A type of Machine Learning where the model learns from data that has been previously labeled by humans. The goal is to learn a function that maps inputs to outputs.
- Example: Train a model with thousands of emails labeled as “spam” or “non-spam” so that it can classify new emails.
Unsupervised Learning
The model learns from data that has not been labeled, searching for hidden patterns or structures on its own.
- Example: Grouping customers with similar purchasing behaviors (clustering) without knowing in advance what those groups are.
Reinforcement Learning (Reinforcement Learning)
A model (called an “agent”) learns to make decisions through trial and error in an environment. Receive rewards for correct actions and penalties for incorrect ones, with the goal of maximizing the total reward.
- Example: Train an AI to play a video game, where it earns points (reward) for advancing in level.
Generative AI (Generative AI)
A branch of AI in which models learn patterns from training data to generate completely new and original content, such as text, images, music or code.
LLM (Large Language Model – Large Language Model)
An advanced type of generative AI model trained on huge amounts of text to understand and generate human language with remarkable fluency and consistency. They are the basis of tools like ChatGPT.
Foundational Models
Large-scale AI models (such as LLMs) trained with a vast amount of data, which can be tailored for a wide range of specific tasks through a process called fine-tuning.
Fine-Tuning (Fine-Tuning)
The process of taking a pre-trained model (such as a foundational model) and retraining it with a smaller, more specific data set to tailor it to a particular task.
Transfer Learning (Transfer Learning)
A technique in which a model developed for one task is reused as a starting point for a model in a second related task. It is an efficient way to take advantage of the knowledge already learned.
Transformers
A neural network architecture that revolutionized natural language processing (NLP). It is especially good at handling sequences of data, such as text, by paying “attention” to the relationships between all the words in the sequence, regardless of their distance. It is the base architecture of most modern LLMs.
Interaction and Development with AI
Key terms about how humans interact with and build on AI systems.
Prompt Engineering
The art and science of designing and refining the instructions (prompts) given to a generative AI model to obtain the most accurate, relevant and high-quality results.
RAG (Retrieval-Augmented Generation – Recovery-Augmented Generation)
A technique that improves the performance of LLMs by allowing them to access external, up-to-date information (such as a database or the Internet) before generating a response. This reduces “hallucinations” and bases answers on verifiable data.
AI Agent
An autonomous system that uses AI to perceive its environment, make decisions, and execute actions to achieve specific goals. Agents can perform complex, multi-step tasks without constant human intervention.
- Example: An agent who can plan an entire trip: find flights, book hotels, and add the itinerary to the calendar, all from a simple request.
Ethics, Security and Governance
Crucial concepts related to the social impact and regulation of AI.
Algorithmic Bias (Bias)
The tendency of an AI system to produce results that are systematically harmful to certain groups or individuals. Generally, it is the result of biases present in the training data or in the design of the algorithm.
Hallucination (Hallucination)
A phenomenon in which a generative AI model produces incorrect, made-up or nonsensical information, but presents it with complete confidence as if it were a real fact.
Explainable AI (XAI – Explainable AI)
A set of methods and techniques that seek to make the decisions and predictions of AI models understandable to humans, opening the “black box” of complex algorithms.
AI Transparency
The principle that the operation, data and decisions of an AI system must be accessible and understandable to its users and regulators.
AI Governance
The set of policies, standards, laws and ethical frameworks that govern the development, deployment and responsible use of artificial intelligence in society.
AI Safety
The measures and practices designed to protect AI systems against attacks, manipulation, or malicious use that could compromise their operation or cause harm.
Sustainable AI
The focus on the design and use of AI systems that minimize their environmental impact, especially in terms of energy consumption and computational resources.
The Future and Avant-garde Concepts
Terms that point to where AI is going.
Quantum Computing and AI
The intersection of quantum computing and AI, which promises to solve optimization and machine learning problems that are intractable for classical computers.
Edge AI
Running AI algorithms directly on local devices (such as phones, sensors, or cars) instead of in the cloud. This reduces latency, saves bandwidth, and improves privacy.
Federated Learning (Federated Learning)
An AI training technique where a model is trained on multiple decentralized devices (such as mobile phones) without the private data of those devices leaving them.
Multi-agent Systems
A system composed of multiple AI agents that interact with each other (collaborating or competing) to solve problems that are beyond the capabilities of a single agent.
A Constantly Evolving Glossary
This glossary is a living document that evolves alongside the fascinating field of Artificial Intelligence. Technology advances by leaps and bounds, and there are always new concepts to explore.
If you think an important term is missing or have any suggestions to improve this resource, feel free to leave a comment! Your contribution is very valuable to keep this guide as complete and up-to-date as possible.
Last update: September 28, 2025.
This post is also available in: