Machine Learning: A branch of AI where machines learn from data without being explicitly programmed. It allows for the identification of patterns in data to predict outcomes. For example, systems that recommend movies based on a user's history.

Deep Learning: A subfield of machine learning based on multi-layered artificial neural networks inspired by the human brain. They are capable of learning very complex representations from data (images, text, etc.).

Algorithm: A set of mathematical or logical instructions that a computer follows to solve a problem. In AI, algorithms process data and adjust a model so that it learns from them.

Data: Information in digital format (numbers, text, images, etc.) used to train AI models. The more numerous, varied, and high-quality the data, the better the AI can learn.

Parameter: An internal numerical value of an AI model that is adjusted during training. It consists of weights and technical biases. In deep neural networks, there are billions of parameters (for example, GPT-3 has 175 billion) that determine how the model responds to inputs.

Artificial Neural Network: A computational model inspired by the brain, consisting of layers of connected "neurons." Each neuron processes input signals and combines them according to parameters (weights) to generate an output. Many layers (deep) allow for modeling complex relationships.

Generative AI: AI technology that creates new content (text, images, sound, video) from learned patterns. LLMs (like ChatGPT) and image models (DALL·E, Stable Diffusion) are examples of generative AI.

Transformer: A neural network architecture specialized in processing sequences (e.g., text). Transformers use self-attention mechanisms to identify relevant parts of the context. They are the foundation of modern language models (GPT, Gemini, etc.).

Language Model: An AI algorithm trained on text to understand and generate language. Based on a given text, it predicts the most likely next word.

Large Language Model (LLM): An extremely large language model (billions of parameters) trained on massive volumes of text. Thanks to its scale, it can perform diverse tasks (summarizing, translating, creating code, etc.) based on simple natural language prompts.

GPT (Generative Pre-trained Transformer): A generative model trained on a massive scale by analyzing billions of texts (books, articles, programming code) to learn language patterns. Invented by Google in 2017, the transformer architecture allows the AI to understand the context of words based on those before and after them, generating a "short-term memory" that enables more efficient language processing.

Natural Language Processing (NLP): A set of techniques for machines to understand and generate human language. It includes tasks such as sentiment analysis, machine translation, entity recognition, or coherent text generation.

Chatbot: A conversational AI program designed to simulate a chat with a user. For example, ChatGPT is an advanced generative AI chatbot; there are also simpler chatbots programmed with fixed responses.

Virtual Assistant: Software (generally with a voice or text interface) that performs tasks or answers questions (e.g., Siri, Alexa, Google Assistant). It uses NLP to interpret user commands and can execute actions (playing music, answering questions, controlling devices).

Expert System: An AI program that uses specialized knowledge (rules or databases from human experts) to solve specific problems. They mimic the reasoning process of a subject matter expert (e.g., medical diagnosis) and can outperform humans in specific decisions thanks to predefined rules.

Big Data: Sets of data so large and complex that they require special tools to process. The rise of Big Data (billions of user records, sensors, etc.) has been key to training modern AI models.

Technical Bias: An additional parameter to the weights that allows for adjusting the activation of a neuron, even when inputs are small or zero. That is, it helps the model shift the result up or down to improve learning.

Voice Recognition: AI technology that converts spoken audio into text and understands voice commands. It uses language models and neural networks to transcribe and understand what we say.

Computer Vision: Areas of AI dedicated to processing images or video to identify or classify them (object detection, facial recognition, autonomous driving). It uses Convolutional Neural Networks (CNN) that analyze an image in small parts, looking for repeating patterns.