AI Glossary
Clear, beginner-friendly definitions for important artificial intelligence, security, governance, cloud infrastructure, machine learning, and AI systems terms.
A
Accuracy
Accuracy measures how often an AI model makes correct predictions compared to the total number of predictions it makes.
Action Item
An Action Item is a specific task or responsibility assigned to an individual or team during or after a meeting.
Activation Function
A mathematical function that helps determine how strongly a neuron should respond to incoming information.
Adversarial Attack
A technique that manipulates inputs to cause an AI system to produce incorrect or unintended results.
AI Abuse
The harmful, inappropriate, or unauthorized use of AI systems.
AI Accountability
AI Accountability is the principle that individuals and organizations remain responsible for the decisions, outcomes, and impacts of artificial intelligence systems.
AI Agent
An AI agent is a system capable of autonomously performing tasks, making decisions, and interacting with tools or environments.
AI Alignment
The effort to ensure AI systems behave in ways that reflect human goals, values, and intentions.
AI Approval Process
An AI Approval Process is the structured workflow of governance, risk, security, and compliance reviews required before an AI system is deployed.
AI Assistant
A software system that uses AI to help users answer questions, complete tasks, and interact with information through natural language.
AI Assurance
AI Assurance is the process of evaluating AI systems to provide confidence that they are trustworthy, secure, compliant, and operating as intended.
AI Assurance Framework
An AI Assurance Framework is a structured set of governance processes, controls, and assessments used to evaluate whether AI systems are trustworthy, secure, compliant, and operating responsibly.
AI Assurance Program
An AI Assurance Program is a structured organizational program that evaluates, monitors, and improves the trustworthiness, security, and governance of AI systems.
AI Audit
A structured review used to evaluate how an AI system is developed, deployed, governed, and monitored.
AI Automation
The use of AI to perform tasks with minimal human intervention.
AI Chatbot
An AI chatbot is a conversational AI system that interacts with users through natural language.
AI Compliance
The process of ensuring AI systems meet applicable laws, regulations, standards, and organizational requirements.
AI Compliance Framework
An AI Compliance Framework is a structured set of controls, processes, and requirements used to ensure AI systems meet regulatory and organizational obligations.
AI Content Generation
The creation of text, images, audio, video, or other content using AI systems.
AI Continuous Assurance
AI Continuous Assurance is the ongoing process of continuously evaluating AI systems through monitoring, testing, audits, and governance activities to ensure they remain trustworthy and compliant.
AI Control Framework
An AI Control Framework is a structured collection of policies, procedures, and safeguards used to reduce AI risks and support responsible AI governance.
AI Controls
AI Controls are policies, procedures, safeguards, and technical measures that help reduce risks and ensure AI systems operate responsibly.
AI Control Testing
AI Control Testing is the process of evaluating whether AI governance controls and safeguards are operating effectively to reduce AI-related risks.
AI Copilot
An AI-powered assistant designed to help users complete specific tasks within a workflow or application.
AI Decision Register
An AI Decision Register is a centralized record of important governance, risk, and operational decisions made throughout the lifecycle of an AI system.
AI Governance
The policies, processes, and controls used to guide how AI systems are developed, deployed, and managed.
AI Governance Charter
An AI Governance Charter is a formal document that defines the purpose, roles, responsibilities, and authority of an organization's AI governance program.
AI Governance Framework
An AI Governance Framework is a structured system of policies, processes, roles, and controls used to manage artificial intelligence responsibly throughout its lifecycle.
AI Governance Maturity Model
An AI Governance Maturity Model is a framework used to measure and improve an organization's AI governance capabilities over time.
AI Guardrails
Rules, controls, and constraints designed to keep AI systems operating safely and appropriately.
AI Hallucination
An AI hallucination occurs when an AI system generates incorrect or fabricated information while presenting it confidently.
AI Incident Management
AI Incident Management is the structured process of identifying, responding to, documenting, investigating, and resolving incidents involving artificial intelligence systems.
AI Infrastructure
The hardware, software, networking, and services required to build, deploy, and operate AI systems.
AI Inventory
An AI Inventory is a centralized record of the artificial intelligence systems an organization develops, deploys, or uses to support governance and oversight.
AI Lifecycle Management
AI Lifecycle Management is the practice of managing AI systems throughout planning, development, deployment, monitoring, maintenance, and retirement.
AI Meeting Assistant
An AI Meeting Assistant is an artificial intelligence tool that automates meeting tasks such as transcription, summaries, note-taking, and action item tracking.
AI Model
An AI model is a trained computational system that learns patterns from data to make predictions or generate outputs.
AI Operating Model
An AI Operating Model defines how an organization structures its people, processes, governance, and technology to manage AI systems effectively.
AI Oversight
AI Oversight is the ongoing process of monitoring, reviewing, and supervising AI systems to ensure they operate safely, responsibly, and in accordance with organizational policies.
AI Policy
An AI Policy is a documented set of rules, principles, and requirements that guide how an organization develops, deploys, and uses artificial intelligence responsibly.
AI Privacy
The practices used to protect personal and sensitive information within AI systems.
AI Productivity
The use of artificial intelligence to help people complete tasks more efficiently and effectively.
AI Reasoning Model
An AI reasoning model is an AI system designed to analyze information and solve complex problems before generating a response.
AI Red Teaming
The practice of actively testing AI systems to identify weaknesses, risks, and vulnerabilities.
AI Review Board
An AI Review Board is a cross-functional group responsible for evaluating significant AI projects to ensure they meet governance, risk, security, and compliance requirements before deployment.
AI Risk Assessment
An AI Risk Assessment is the process of identifying, evaluating, and documenting potential risks associated with an AI system.
AI Risk Management
The process of identifying, assessing, and reducing risks associated with AI systems.
AI Safety
The field focused on reducing risks and unintended consequences associated with AI systems.
AI Security
The practice of protecting AI systems, models, data, and infrastructure from threats and misuse.
AI Stewardship
AI Stewardship is the responsible management of artificial intelligence systems to maximize their benefits while protecting people, organizations, and society.
AI System Inventory
An AI System Inventory is a detailed catalog of individual AI systems that records technical, governance, operational, and risk-related information for each AI deployment.
AI System Owner
An AI System Owner is the individual responsible for overseeing an AI system throughout its lifecycle, including governance, performance, compliance, and ongoing operation.
AI Threat Model
A structured approach used to identify, evaluate, and prioritize risks affecting AI systems.
AI Transparency
The practice of providing clear information about how AI systems operate, use data, and generate outputs.
AI Workflow
A sequence of tasks that incorporates AI systems to help complete a larger process.
API
A set of rules that allows different software systems to communicate with each other.
Artificial Intelligence
Artificial Intelligence (AI) is the field of computer science focused on creating systems that can perform tasks that normally require human intelligence.
Artificial Neuron
A mathematical unit within a neural network that receives inputs, processes information, and produces an output.
Artificial Superintelligence
Artificial Superintelligence refers to a hypothetical future AI system that would surpass human intelligence across virtually all cognitive tasks.
Attention Mechanism
A technique that helps AI models focus on the most relevant information when processing data.
Autonomous System
An autonomous system is a system that can perform tasks, make decisions, or take actions with limited or no direct human intervention.
B
Backpropagation
A learning process that helps a neural network identify and correct mistakes during training.
Batch Size
The number of training examples processed together before a model updates its internal parameters.
Bias-Variance Tradeoff
The bias-variance tradeoff describes the balance between models that are too simple and models that are too complex.
Big Data
Big data refers to extremely large and complex datasets that can be analyzed to discover patterns, trends, and insights.
C
Chain-of-Thought Prompting
A prompting technique that encourages an AI model to reason through a problem step by step before providing an answer.
Classification
Classification is a machine learning task where an AI system places information into predefined categories based on patterns learned from data.
Cloud Computing
The delivery of computing resources such as storage, networking, and processing power over the internet.
Clustering
Clustering is a machine learning technique that groups similar data together by identifying natural patterns and relationships within a dataset.
Computer Vision
Computer Vision is the field of AI focused on enabling computers to analyze, interpret, and understand visual information.
Confusion Matrix
A table that summarizes a model's correct and incorrect predictions.
Consent
Consent is the voluntary permission given by an individual for their data to be collected, processed, or used for a specific purpose.
Context Window
The amount of information an AI model can consider at one time when generating a response.
Continuous Monitoring
Continuous Monitoring is the ongoing process of tracking AI systems, risks, controls, and performance to ensure they continue operating responsibly.
Conversation Intelligence
Conversation Intelligence is the use of artificial intelligence to analyze conversations and extract insights that improve communication, collaboration, and decision-making.
Convolutional Neural Network (CNN)
A neural network architecture commonly used for image recognition and computer vision tasks.
Cross Validation
Cross validation is a model evaluation technique that repeatedly trains and tests a model using different portions of a dataset to produce more reliable performance measurements.
Cybersecurity
The practice of protecting systems, networks, applications, and information from threats and attacks.
D
Data Governance
Data Governance is the framework of policies, processes, and responsibilities used to manage data throughout its lifecycle
Data Lake
A centralized storage system that holds large amounts of raw data in its original format.
Data Mining
Data mining is the process of discovering patterns, trends, and useful information within large datasets.
Data Pipeline
A system that collects, transforms, and moves data between different locations and applications.
Data Poisoning
The manipulation of training data to influence how an AI model learns and behaves.
Data Science
Data science is the practice of collecting, analyzing, and interpreting data to generate insights and support decision-making.
Dataset
A dataset is a collection of information used to train, test, or evaluate an AI system. It provides the examples that help a model learn patterns and make predictions.
Data Warehouse
A centralized system that stores structured and organized data for reporting, analysis, and business intelligence.
Decision Tree
A decision tree is a machine learning model that makes predictions by following a series of questions and branching decisions that lead to an outcome.
Deepfake
AI-generated or AI-manipulated content designed to realistically imitate a person's appearance, voice, or behavior.
Deep Learning
Deep Learning is a specialized area of machine learning that uses large neural networks to process complex data and power modern AI systems.
Differential Privacy
A privacy-preserving technique designed to reduce the risk of identifying individuals within a dataset.
Diffusion Model
A type of generative AI model that creates content by gradually transforming random noise into meaningful outputs.
Digital Sovereignty
Digital Sovereignty refers to an organization's or nation's ability to control its digital data, technologies, and infrastructure.
Dimensionality Reduction
A technique used to reduce the number of variables in a dataset while preserving important information.
Distributed Computing
A computing approach that uses multiple computers working together to complete tasks.
E
Edge AI
AI systems that run directly on local devices rather than relying entirely on cloud-based processing.
Embedding
A numerical representation that helps AI models understand relationships between words, concepts, or objects.
Embedding Model
An AI model that converts information into embeddings so that relationships between concepts can be understood mathematically.
Encryption
A security technique that converts information into a protected format that can only be accessed with proper authorization.
Ensemble Learning
A machine learning approach that combines multiple models to improve prediction performance.
Epoch
One complete pass through the entire training dataset during model training.
Ethics
AI Ethics is the application of moral principles that guide the responsible development, deployment, and use of artificial intelligence.
ETL
A process that extracts, transforms, and loads data between systems.
Expert Systems
An expert system is a computer program designed to make decisions or provide recommendations using predefined rules and knowledge from human experts.
Explainable AI
An approach that helps people understand how AI systems generate outputs, recommendations, or decisions.
F
F1 Score
A metric that combines Precision and Recall into a single measurement of model performance.
Feature Engineering
The process of selecting, transforming, or creating data features that help an AI model learn more effectively.
Features
Features are the individual pieces of information within a dataset that help an AI model learn patterns and make predictions.
Federated Learning
A machine learning approach that allows models to learn from distributed data without requiring all data to be collected in one location.
Few-Shot Learning
The ability of an AI model to learn or perform a task after being given only a small number of examples.
Fine-Tuning
Fine-tuning is the process of further training an existing AI model on specialized data to improve performance for a specific task.
Foundation Model
A foundation model is a large AI model trained on broad datasets that can be adapted for many different applications.
G
General AI
General AI is a hypothetical form of artificial intelligence capable of performing a wide variety of intellectual tasks at a level similar to human intelligence.
Generative Adversarial Network (GAN)
An AI architecture that uses two competing neural networks to generate realistic content such as images and videos.
Generative AI
Generative AI refers to AI systems capable of creating new content such as text, images, code, audio, and video.
Governance Committee
A Governance Committee is a cross-functional group responsible for overseeing AI governance, reviewing AI initiatives, and ensuring AI systems are managed responsibly.
GPU
A processor designed to perform many calculations simultaneously, making it useful for AI and machine learning workloads.
Gradient Boosting
Gradient Boosting is a machine learning technique where new models are built to correct the errors made by previous models. Over time, the combined system becomes more accurate.
Gradient Descent
An optimization technique that helps a model gradually reduce errors and improve predictions.
H
Hidden Layer
A layer within a neural network that processes information between the input and output layers.
Human Oversight
Human Oversight is the practice of ensuring people remain involved in monitoring, reviewing, or approving important AI decisions.
Hyperparameter
A hyperparameter is a setting chosen before model training begins that influences how an AI model learns from data.
I
Identity and Access Management (IAM)
The processes and technologies used to control who can access systems and what actions they can perform.
Inference
The process of using a trained AI model to generate predictions, classifications, or outputs.
Inference Server
A system that hosts AI models and processes requests from users or applications.
Input Layer
The first layer of a neural network that receives data before it is processed by other layers.
J
K
Key Risk Indicator (KRI)
A measurable metric used to monitor changes in risk over time.
K-Nearest Neighbors (KNN)
K-Nearest Neighbors is a machine learning algorithm that makes predictions based on the most similar examples in a dataset.
Knowledge Management
Knowledge Management is the process of organizing, storing, sharing, and maintaining information so it can be easily accessed and used across an organization.
Knowledge Representation
Knowledge representation is the process of organizing information in a structured way so that a computer system can use, interpret, and reason about it.
Kubernetes
An open-source platform used to automate the deployment, scaling, and management of applications.
L
Labels
Labels are the correct answers or outcomes associated with training data. They help supervised machine learning models learn what they are trying to predict.
Large Language Model
A Large Language Model (LLM) is an AI system trained on massive amounts of text data to understand and generate human language.
Latency
The amount of time it takes for a system to respond to a request.
Linear Regression
Linear Regression is a machine learning model used to predict numerical values by identifying relationships between variables.
Logistic Regression
Logistic Regression is a machine learning model used for classification tasks, helping determine the probability that something belongs to a particular category.
Long Short-Term Memory (LSTM)
A type of recurrent neural network designed to remember important information over longer periods of time.
M
Machine Learning
Machine Learning (ML) is a branch of AI that enables systems to learn patterns from data and improve performance without being explicitly programmed for every task.
Meeting Analytics
Meeting Analytics is the use of artificial intelligence to analyze meeting data, participation, communication, and collaboration patterns to improve productivity.
Meeting Automation
Meeting Automation is the use of artificial intelligence to automate meeting tasks such as scheduling, transcription, summaries, action items, and follow-up activities.
Meeting Summary
A Meeting Summary is an AI-generated overview of the key topics, decisions, and action items discussed during a meeting.
Meeting Transcription
Meeting Transcription is the process of converting spoken conversations into searchable written text using artificial intelligence or speech recognition technology.
MLOps
A set of practices used to deploy, monitor, maintain, and improve machine learning systems in production.
Model Deployment
The process of making a trained AI model available for real-world use.
Model Documentation
Model Documentation is the collection of records and information that describe how an AI model was developed, tested, deployed, and monitored.
Model Drift
The gradual decline in model performance caused by changes in data, user behavior, or real-world conditions.
Model Evaluation
Model evaluation is the process of measuring how well an AI model performs on data it has not previously seen.
Model Integrity
The assurance that an AI model remains accurate, trustworthy, and free from unauthorized modification.
Model Monitoring
Model Monitoring is the continuous process of tracking an AI model's performance, accuracy, reliability, and behavior after deployment.
Model Poisoning
An attack that attempts to alter the behavior of an AI model by manipulating the model itself during training or updating.
Model Theft
The unauthorized copying, extraction, or acquisition of an AI model.
Model Training
Model training is the process of teaching an AI system to recognize patterns and relationships within data so it can make predictions or decisions.
Multimodal AI
Multimodal AI refers to AI systems that can process and understand multiple forms of information such as text, images, audio, and video.
N
Narrow AI
Narrow AI refers to artificial intelligence systems designed to perform specific tasks rather than possessing broad human-like intelligence.
Natural Language Processing
Natural Language Processing (NLP) is the field of AI focused on helping computers understand, interpret, and generate human language.
Neural Network
A neural network is a computational model inspired by the human brain that helps AI systems recognize patterns and process information.
O
Open Source Model
An open source model is an AI model that is publicly available for inspection, modification, and deployment.
Output Layer
The final layer of a neural network that produces the model's prediction, classification, or output.
Overfitting
Overfitting occurs when a machine learning model learns the training data too closely, causing it to perform poorly on new, unseen data.
P
Pattern Recognition
Pattern recognition is the ability of an AI system to identify relationships, trends, or recurring structures within data. It is one of the foundational capabilities that allows modern AI systems to make predictions and generate useful outputs.
Phishing
A type of social engineering attack that attempts to trick people into revealing sensitive information.
Positional Encoding
Information added to tokens that helps a model understand their order within a sequence.
Precision
Measures how many positive predictions made by a model were actually correct.
Predictive Analytics
Predictive analytics uses historical data, statistics, and AI models to forecast future outcomes.
Principal Component Analysis (PCA)
A dimensionality reduction method that transforms data into a smaller set of components that capture most of the important patterns.
Privacy by Design
Privacy by Design is the practice of building privacy protections into systems and processes from the beginning rather than adding them later.
Prompt Engineering
Prompt Engineering is the practice of designing effective instructions and inputs to improve AI-generated responses and outputs.
Prompt Injection
An attack that attempts to manipulate an AI system by providing instructions that interfere with its intended behavior.
Proprietary Model
A proprietary model is an AI model that is owned and controlled by a company or organization.
R
Random Forest
A Random Forest is a machine learning model that combines many decision trees to make more accurate and reliable predictions. It reduces the risk of relying on a single tree's mistakes.
Recall
Measures how many actual positive cases were successfully identified by a model.
Recurrent Neural Network (RNN)
A neural network architecture designed to process sequential information such as text, speech, and time-series data.
Regression
Regression is a machine learning task that predicts numerical values, such as prices, sales, or future outcomes, based on patterns in data.
Reinforcement Learning
Reinforcement learning is a type of machine learning where an AI system learns through trial and error by receiving rewards or penalties for its actions.
Responsible AI
Responsible AI refers to the ethical, secure, transparent, and accountable development and use of artificial intelligence systems.
Retrieval-Augmented Generation (RAG)
A technique that combines information retrieval with AI generation to produce more accurate and context-aware responses.
Risk Appetite
The overall amount of risk an organization is willing to accept while pursuing its objectives.
Risk Assessment
The process of identifying potential risks and evaluating their likelihood and potential impact.
Risk Mitigation
The process of reducing the likelihood or impact of identified risks.
Risk Register
A centralized document used to record, monitor, and manage identified risks.
Risk Tolerance
The acceptable level of variation or deviation an organization is willing to allow for a specific risk.
Robotics
Robotics is a field of technology that combines software, sensors, and machines to perform tasks automatically or with limited human intervention.
ROC Curve
A graph that shows how well a model distinguishes between positive and negative outcomes across different thresholds.
S
Scalability
The ability of a system to handle increasing amounts of work without significant performance problems.
Secure AI Development
The practice of incorporating security principles throughout the AI development lifecycle.
Self-Attention
A type of attention mechanism that allows a model to examine relationships between different parts of the same input.
Self-Supervised Learning
Self-supervised learning is a machine learning technique where AI creates its own learning tasks from unlabeled data.
Semi-Supervised Learning
Semi-supervised learning combines a small amount of labeled data with a larger amount of unlabeled data to improve model performance.
Social Engineering
A technique that manipulates people into revealing information or performing actions that benefit an attacker.
Speech Recognition
Speech Recognition is the artificial intelligence technology that converts spoken language into written text.
Stochastic Gradient Descent (SGD)
A variation of Gradient Descent that updates a model using smaller subsets of training data.
Structured Data
Data that is organized into a predefined format, making it easy to store, search, and analyze.
Supervised Learning
Supervised learning is a machine learning method where models learn from labeled examples containing both inputs and correct answers.
Supply Chain Security
The practice of protecting the software, models, datasets, and dependencies used within AI systems.
Support Vector Machine (SVM)
A Support Vector Machine is a machine learning model that separates data into categories by finding the clearest possible boundary between groups.
Symbolic AI
Symbolic AI is an approach to artificial intelligence that relies on explicit rules, logic, and structured knowledge rather than learning patterns from large datasets.
Synthetic Data
Artificially generated data that mimics real-world data without being collected from actual events or individuals.
Synthetic Media
Content that is generated or significantly modified using artificial intelligence.
T
Temperature
A setting that controls how predictable or creative an AI model's outputs are.
Text-to-Image Generation
Text-to-image generation is the process of creating images from written descriptions using AI.
Text-to-Video Generation
Text-to-video generation is the process of creating videos from written instructions using AI.
Token
A small unit of text processed by an AI model, often representing a word, part of a word, or punctuation.
Tokenization
The process of breaking text into smaller units called tokens before it is processed by an AI model.
TPU
A specialized processor designed specifically for machine learning and AI applications.
Training Data
Training data is the information used to teach AI models how to recognize patterns and perform tasks.
Transfer Learning
Transfer learning is an AI technique where knowledge learned from one task is reused to help solve another related task.
Transformer Model
A transformer model is a deep learning architecture that enables modern AI systems to process and understand language efficiently.
U
Underfitting
Underfitting occurs when an AI model fails to learn enough from training data, causing poor performance on both known and unseen data.
Unstructured Data
Data that does not follow a predefined format, such as documents, emails, images, videos, and audio files.
Unsupervised Learning
Unsupervised learning is a machine learning method where AI systems identify patterns in data without being given correct answers.