AI Glossary 2026

The process of integrating AI technologies into business operations, requiring strategy, change management, and skill development.

The challenge of ensuring AI systems act in accordance with human values, intentions, and ethical principles.

Systematic unfairness in AI systems caused by biased training data, flawed algorithms, or discriminatory design choices.

The study of moral principles and guidelines for developing and deploying AI systems responsibly and beneficially.

Frameworks, policies, and regulations for overseeing AI development and deployment to ensure safety and accountability.

Research and practices ensuring AI systems operate safely, without causing unintended harm to humans or society.

A business approach that prioritizes AI integration across all operations, products, and decision-making processes.

A set of step-by-step instructions or rules that a computer follows to solve a problem or complete a task. AI algorithms enable machines to learn patterns from data.

A set of protocols allowing different software applications to communicate, enabling integration of AI capabilities into products.

A type of AI that can understand, learn, and apply knowledge across a wide range of tasks at a human level. Unlike narrow AI, AGI would possess general cognitive abilities.

The simulation of human intelligence processes by computer systems, including learning, reasoning, problem-solving, perception, and language understanding.

A hypothetical AI that surpasses human intelligence in virtually all domains, including creativity, problem-solving, and social intelligence.

A technique that allows models to focus on relevant parts of input when producing output, crucial for tasks like translation and summarization.

The use of technology to perform tasks with minimal human intervention. AI-powered automation can handle complex decision-making processes.

Automated Machine Learning tools that automatically select, configure, and train machine learning models with minimal human input.

A training algorithm for neural networks that calculates gradients and adjusts weights by propagating errors backward through the network layers.

The number of training samples processed before updating the model parameters. Larger batches provide stable gradients but require more memory.

Bidirectional Encoder Representations from Transformers - a pre-trained language model that understands context from both directions for NLP tasks.

Systematic errors in AI predictions caused by flawed assumptions in the learning algorithm or training data that can lead to unfair outcomes.

AI systems whose internal decision-making processes are opaque and difficult to interpret or explain to humans.

An AI application that simulates human conversation through text or voice, using NLP to understand and respond to user queries.

OpenAI's conversational AI model that can engage in human-like dialogue, answer questions, and assist with various tasks.

A supervised learning task where the model predicts discrete categories or labels for input data, such as spam detection or image recognition.

An AI assistant developed by Anthropic, designed to be helpful, harmless, and honest, with strong reasoning and analysis capabilities.

An unsupervised learning technique that groups similar data points together based on their features without predefined labels.

A field of AI enabling computers to interpret and understand visual information from images, videos, and the real world.

The maximum amount of text a language model can process at once, measured in tokens. Larger windows allow understanding of longer documents.

AI technologies enabling natural language interactions between humans and machines, powering chatbots and virtual assistants.

A deep learning architecture designed for processing grid-like data such as images, using convolutional layers to detect patterns and features.

AI-powered coding assistant by GitHub/Microsoft that suggests code completions, functions, and solutions based on context.

A technique to evaluate model performance by dividing data into multiple subsets, training on some and testing on others to prevent overfitting.

AI-powered unified view of customer data across touchpoints, enabling personalized experiences and predictive analytics.

OpenAI's AI system that creates realistic images and art from natural language descriptions, revolutionizing digital creativity.

Techniques to artificially increase training data size by creating modified versions of existing data through transformations like rotation or cropping.

A supervised learning algorithm that makes decisions by splitting data based on feature values, creating a tree-like structure of if-then rules.

A subset of machine learning using neural networks with multiple layers to learn complex patterns from large amounts of data automatically.

AI-generated synthetic media that convincingly replaces one person's likeness with another, raising concerns about misinformation.

A generative AI model that creates images by gradually removing noise from random data, guided by learned patterns from training.

A virtual replica of physical assets, processes, or systems that uses AI to simulate, predict, and optimize real-world performance.

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

A dense vector representation of words, sentences, or documents that captures semantic meaning in a numerical format for machine processing.

A technique that combines multiple models to produce better predictions than any single model, including methods like bagging and boosting.

An NLP task that identifies and classifies named entities in text, such as people, organizations, locations, and dates.

One complete pass through the entire training dataset during model training. Multiple epochs are typically needed for convergence.

The degree to which AI decisions can be understood by humans, essential for trust and accountability in high-stakes applications.

AI technology that identifies or verifies individuals by analyzing facial features and patterns in images or video.

The principle that AI systems should make unbiased decisions that treat all individuals and groups equitably.

The process of creating, selecting, and transforming input variables to improve machine learning model performance.

The process of automatically or manually identifying relevant characteristics from raw data that can be used for machine learning.

The ability of a model to learn new tasks from only a few examples, without extensive retraining on large datasets.

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

A neural network architecture with two competing networks—a generator creating data and a discriminator evaluating it—producing realistic outputs.

AI systems that create new content—text, images, code, music—rather than just analyzing or classifying existing data.

A family of large language models by OpenAI that generate human-like text by predicting the next token in a sequence.

An optimization algorithm that iteratively adjusts model parameters to minimize the loss function by moving in the direction of steepest descent.

When AI models generate plausible-sounding but factually incorrect or fabricated information that was not in their training data.

A platform and community for sharing machine learning models, datasets, and tools, democratizing access to AI technology.

The combination of AI, machine learning, and RPA to automate complex business processes end-to-end with minimal human intervention.

Configuration settings that control the learning process, such as learning rate or number of layers, set before training begins.

A computer vision task that assigns labels or categories to entire images based on their visual content.

AI techniques that create new images from text descriptions, sketches, or other inputs, using models like diffusion or GANs.

Dividing an image into distinct regions or segments, identifying pixel-level boundaries of objects and areas.

AI-powered extraction and processing of data from documents like invoices, contracts, and forms with high accuracy.

The process of identifying the purpose or goal behind a user's text input, essential for chatbots and virtual assistants.

The ability to understand how an AI model makes predictions, crucial for debugging and building user trust.

An interactive computing environment for creating and sharing documents containing live code, visualizations, and narrative text.

An unsupervised learning algorithm that partitions data into K clusters by minimizing the distance between points and their cluster centers.

A framework for developing applications powered by language models, simplifying chains of prompts, memory, and tool use.

AI models trained on massive text datasets that can understand and generate human-like text, powering applications like ChatGPT and Claude.

A hyperparameter that controls how much to adjust model weights during training. Too high causes instability; too low slows convergence.

An NLP technique that reduces words to their base dictionary form (lemma), considering context and part of speech.

A mathematical function that measures the difference between predicted and actual values, guiding the optimization of model parameters.

A type of recurrent neural network designed to learn long-term dependencies, using gates to control information flow over time.

A subset of AI where systems learn and improve from experience without being explicitly programmed, using algorithms to find patterns in data.

AI-powered automatic translation of text or speech from one language to another, using neural networks for context-aware results.

An AI image generation tool that creates artistic visuals from text prompts, known for distinctive aesthetic quality.

Practices combining machine learning, DevOps, and data engineering to deploy and maintain ML models in production reliably.

A mathematical representation learned from training data that can make predictions or decisions on new, unseen data.

An NLP technique that identifies and classifies named entities like people, places, and organizations in unstructured text.

AI technology that converts structured data into human-readable text, used in report generation and content creation.

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

A subset of NLP focused on comprehending the meaning and intent behind human language, beyond just parsing text.

A computing system inspired by biological neural networks, consisting of interconnected nodes (neurons) organized in layers that process information.

AI platforms enabling users to build and deploy machine learning models without programming knowledge through visual interfaces.

A computer vision task that identifies and locates objects within images or video, drawing bounding boxes around them.

Technology that converts images of text into machine-readable text, enabling digitization of printed documents.

An AI research company that developed GPT, ChatGPT, and DALL-E, pioneering large language models and AI safety research.

When a model learns training data too well, including noise and outliers, resulting in poor performance on new, unseen data.

The simplest type of artificial neural network, consisting of a single neuron that makes binary classifications based on weighted inputs.

Computer vision technique that detects human body positions and movements by identifying key body joint locations.

The initial training phase where a model learns general features from a large dataset before being fine-tuned for specific tasks.

Using AI and statistical techniques to analyze historical data and predict future outcomes, trends, and behaviors.

AI-driven analysis of event logs to discover, monitor, and improve business processes based on actual execution data.

The input text or instructions given to an AI model to guide its response or generation, crucial for directing LLM outputs.

The practice of designing and optimizing prompts to elicit desired responses from AI models, improving output quality and relevance.

An open-source machine learning framework popular for research and production, known for dynamic computation graphs and Python integration.

A technique combining retrieval systems with generative models to provide accurate, up-to-date responses grounded in external knowledge.

An ensemble learning method that builds multiple decision trees and combines their predictions for improved accuracy and robustness.

AI systems that suggest relevant content, products, or actions to users based on their behavior and preferences.

A neural network architecture designed for sequential data, where connections form cycles allowing information to persist across time steps.

A supervised learning task where the model predicts continuous numerical values, such as price prediction or temperature forecasting.

Techniques to prevent overfitting by adding penalties to the loss function, such as L1 (Lasso) or L2 (Ridge) regularization.

A learning paradigm where agents learn optimal actions through trial and error, receiving rewards or penalties based on their decisions.

Developing AI with consideration for ethics, fairness, transparency, privacy, and societal impact throughout its lifecycle.

Software robots that automate repetitive, rule-based tasks by mimicking human interactions with digital systems.

A mechanism allowing models to weigh the importance of different parts of an input sequence relative to each other for better context understanding.

Search technology that understands the meaning and intent behind queries, not just keyword matching, for more relevant results.

An NLP technique that determines the emotional tone or opinion expressed in text, classifying it as positive, negative, or neutral.

An open-source deep learning model that generates detailed images from text descriptions using latent diffusion techniques.

An NLP technique that reduces words to their root form by removing suffixes, simpler but less accurate than lemmatization.

Common words like "the," "is," and "at" that are often filtered out in NLP preprocessing as they carry little semantic meaning.

A machine learning approach where models learn from labeled training data, with known input-output pairs guiding the learning process.

Google's open-source machine learning framework for building and deploying ML models at scale, with extensive ecosystem support.

An NLP task that assigns predefined categories or labels to text documents, used in spam detection and topic categorization.

AI-powered condensation of long documents into shorter versions while preserving key information and meaning.

The basic unit of text processed by language models, which can be a word, part of a word, or punctuation mark.

The process of breaking text into smaller units (tokens) for processing by NLP models, essential for text analysis.

A technique where a model trained on one task is repurposed for a different but related task, reducing training time and data requirements.

A neural network architecture using self-attention mechanisms to process sequential data in parallel, forming the basis of modern language models.

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

A machine learning approach where models find patterns in unlabeled data without predefined outputs, discovering hidden structures.

The sensitivity of a model to fluctuations in training data. High variance leads to overfitting; low variance may cause underfitting.

A database optimized for storing and querying high-dimensional vectors (embeddings), enabling fast similarity search for AI applications.

AI-powered software that performs tasks or services based on voice or text commands, like Siri, Alexa, or Google Assistant.

Numerical parameters in neural networks that are adjusted during training to minimize prediction errors and improve model accuracy.

A neural network model that learns word embeddings by predicting words from context, capturing semantic relationships in vector space.

AI systems designed to provide clear explanations of their decisions and predictions in human-understandable terms.

A real-time object detection algorithm that processes images in a single pass, making it fast enough for video applications.

The ability of a model to perform tasks it was not explicitly trained on, using general knowledge to handle new categories.