AI Glossary
Key AI and enterprise technology terms — practical, jargon-free explanations.
135 terms
A
A/B Testing AI Models
A/B testing for AI models compares multiple model versions in production to determine which delivers better business outcomes with statistical confidence.
Read more →A2A (Agent-to-Agent Protocol)
Protocol for communication between AI agents from different vendors — enabling collaboration between Google, Microsoft, Salesforce agents.
Read more →Adversarial Attacks on AI
Adversarial attacks exploit vulnerabilities in AI models by crafting inputs designed to cause misclassification or unexpected behavior.
Read more →Agentic AI
AI systems capable of autonomous planning, decision-making, and executing multi-step tasks without constant human oversight.
Read more →AI Act Risk Classification
The EU AI Act classifies AI systems into four risk levels — unacceptable, high, limited, and minimal — each with specific regulatory requirements.
Read more →AI Alignment
The challenge of ensuring AI systems behave in accordance with human values, intentions, and safety requirements.
Read more →AI and GDPR
GDPR compliance for AI systems requires careful handling of personal data throughout the machine learning lifecycle, from training to inference.
Read more →AI as a Service (AIaaS)
Cloud-based AI services that allow organizations to access artificial intelligence capabilities without building infrastructure from scratch.
Read more →AI Audit
Systematic assessment of AI systems for security, regulatory compliance, result quality, and business risk.
Read more →AI Benchmarks
AI benchmarks are standardized evaluation frameworks that measure and compare the capabilities of AI models across specific tasks and domains.
Read more →AI Bias
Systematic prejudices in AI model outputs resulting from unequal training data — discrimination risk and regulatory non-compliance.
Read more →AI Center of Excellence
A dedicated organizational unit that drives AI adoption by providing expertise, standards, best practices, and shared resources.
Read more →AI Code Generation
Using AI models to automatically write, complete, and transform source code based on natural language instructions or context.
Read more →AI Compliance Testing
AI compliance testing systematically verifies that AI systems meet regulatory, ethical, and organizational requirements before and during deployment.
Read more →AI Data Anonymization
Automatically removing or masking personal data (PII) in training sets and AI model queries, GDPR-compliant.
Read more →AI Document Summarization
Using AI to automatically condense lengthy documents into concise summaries while preserving key information and context.
Read more →AI Ethics
AI ethics examines the moral principles and societal implications of artificial intelligence, guiding organizations toward beneficial and fair AI development.
Read more →AI Governance
Organizational framework for managing AI in the enterprise — policies, processes, accountability, and regulatory compliance.
Read more →AI Guardrails
Protective mechanisms limiting AI model behavior — content filters, output validation, permission limits, and security controls.
Read more →AI Image Generation
Creating original images from text descriptions or other inputs using AI models like diffusion networks and GANs.
Read more →AI Implementation Roadmap
A phased plan for deploying AI in an organization, covering assessment, pilots, scaling, and enterprise-wide integration.
Read more →AI in Customer Service
How artificial intelligence transforms customer support through intelligent automation, personalization, and 24/7 availability.
Read more →AI in Finance
How AI is transforming financial services through fraud detection, risk assessment, trading automation, and regulatory compliance.
Read more →AI in Healthcare
How AI is advancing medical diagnosis, drug discovery, patient care, and health system efficiency while navigating regulatory requirements.
Read more →AI in HR and Recruitment
Applications of AI in human resources, from candidate screening to employee engagement, with attention to bias and regulatory risks.
Read more →AI in Legal Industry
How AI is transforming legal work through document analysis, contract review, legal research, and compliance automation.
Read more →AI in Logistics
How AI optimizes supply chains, route planning, demand forecasting, and warehouse operations for greater efficiency.
Read more →AI in Manufacturing
How AI optimizes manufacturing through quality control, predictive maintenance, process optimization, and smart factory automation.
Read more →AI in Marketing
How AI transforms marketing through personalization, content generation, audience targeting, and campaign optimization.
Read more →AI in Software Testing
Applying AI to automate test creation, execution, and maintenance, improving coverage and catching defects earlier in development.
Read more →AI Inference
The process of generating responses by a trained AI model — the production stage where the model processes inputs and returns results.
Read more →AI Integration with IT Systems
Connecting AI capabilities with existing enterprise IT infrastructure, from ERPs and CRMs to databases and communication platforms.
Read more →AI Literacy
Mandatory from February 2025 — the ability to understand and responsibly use AI, required by AI Act Article 4.
Read more →AI Maturity Model
A structured framework for assessing an organization's readiness, capabilities, and progression in adopting artificial intelligence.
Read more →AI Model Evaluation
AI model evaluation systematically assesses model performance using metrics, test datasets, and domain-specific criteria to ensure production readiness.
Read more →AI Model Monitoring
AI model monitoring continuously tracks model performance, data quality, and system health in production to detect degradation and ensure reliable AI operations.
Read more →AI Observability
Real-time monitoring of AI systems — tracking performance, costs, response quality, and anomalies in production deployments.
Read more →AI Orchestration
Coordinating multiple AI models and agents working together on complex tasks — from resource allocation to data flow management.
Read more →AI Pair Programming
Collaborating with an AI assistant during software development for real-time code suggestions, debugging, and problem-solving.
Read more →AI Pipeline
An AI pipeline is an automated sequence of data processing, model training, evaluation, and deployment steps that produces production-ready AI systems.
Read more →AI Process Automation
Using artificial intelligence to automate complex business processes that involve judgment, unstructured data, and dynamic decision-making.
Read more →AI Procurement
The process of evaluating, selecting, and acquiring AI solutions, requiring specialized criteria beyond traditional IT procurement.
Read more →AI Reasoning
The ability of AI systems to perform logical thinking, multi-step problem solving, and structured analysis beyond pattern matching.
Read more →AI Red Teaming
Testing AI system security through simulated attacks — finding vulnerabilities, guardrail bypasses, and model manipulation methods.
Read more →AI Response Streaming
A technique for delivering AI model outputs incrementally as they are generated, reducing perceived latency.
Read more →AI Sandbox
An isolated environment for safely experimenting with AI models, testing new approaches, and validating solutions before production deployment.
Read more →AI Supply Chain Security
AI supply chain security addresses risks from third-party models, datasets, libraries, and infrastructure used in enterprise AI systems.
Read more →AI Tokenization
Process of converting text into tokens (word/character fragments) understood by the AI model — directly impacts costs and quality.
Read more →AI Video Generation
Using AI to create, edit, and enhance video content from text prompts, images, or existing footage with minimal manual production.
Read more →AI Watermarking
AI watermarking embeds detectable signals in AI-generated content to enable provenance tracking and authenticity verification.
Read more →AI-Powered Knowledge Management
Using AI to capture, organize, retrieve, and generate organizational knowledge, making institutional expertise accessible at scale.
Read more →AI-Powered OCR
Advanced optical character recognition enhanced by AI to accurately extract text from diverse documents, handwriting, and images.
Read more →Attention Mechanism
A neural network technique that allows models to focus on the most relevant parts of input data when producing outputs.
Read more →Autonomous AI Agents
AI systems that independently plan, execute, and adapt sequences of actions to accomplish complex goals with minimal human intervention.
Read more →C
Chain of Thought
Prompting technique where the AI model "thinks aloud" — reasoning step by step, improving accuracy on complex questions.
Read more →Chatbot vs AI Agent
Understanding the fundamental differences between simple conversational chatbots and autonomous AI agents capable of independent action.
Read more →CI/CD for AI
CI/CD for AI extends continuous integration and delivery practices to machine learning, automating testing, validation, and deployment of models and data pipelines.
Read more →Cloud AI vs On-Premise AI
Comparing cloud-hosted and on-premise AI deployment models in terms of cost, control, security, scalability, and compliance.
Read more →Computer Use (AI)
AI models' ability to directly control a computer — clicking, typing, navigating interfaces like a human.
Read more →Computer Vision
AI technology that enables machines to interpret and analyze visual information from images, video, and real-world environments.
Read more →Confidential Computing
Confidential computing protects AI data and models during processing using hardware-based trusted execution environments.
Read more →Context Window
Maximum amount of text (tokens) an AI model can process in a single query — a key LLM performance constraint.
Read more →Conversational AI
AI systems that enable natural language dialogue between humans and machines across text and voice channels.
Read more →Corporate AI Strategy
A comprehensive plan that aligns AI initiatives with business objectives, covering technology, talent, data, governance, and culture.
Read more →D
Data Annotation (Data Labeling)
Data annotation is the process of labeling raw data with meaningful tags to create training datasets for supervised machine learning models.
Read more →Data Drift
Data drift occurs when the statistical properties of production data diverge from training data, causing AI model performance to degrade over time.
Read more →Data Poisoning
Data poisoning attacks corrupt AI training datasets to manipulate model behavior, introducing biases or backdoors that persist after training.
Read more →Deepfake Detection
Deepfakes are AI-generated synthetic media that convincingly replicate real people, posing serious risks to enterprise security and trust.
Read more →Differential Privacy
Differential privacy is a mathematical framework that enables AI systems to learn from datasets while providing formal guarantees about individual data protection.
Read more →Digital Twin
A virtual replica of a physical system, process, or asset that uses real-time data and AI for simulation and optimization.
Read more →Document Chunking
The process of splitting documents into smaller, meaningful segments optimized for AI retrieval and processing in RAG systems.
Read more →E
Edge AI
Running AI models directly on end devices — without sending data to the cloud, with minimal latency.
Read more →Embedding (Vector Representation)
Representing text, images, or audio as number vectors — the foundation of semantic search and RAG systems.
Read more →Emergent Abilities in AI
Capabilities that unexpectedly arise in large AI models at certain scales, not present in smaller versions of the same architecture.
Read more →EU AI Act Guide
The EU AI Act is the world's first comprehensive legal framework for artificial intelligence, establishing rules based on risk levels.
Read more →Explainable AI (XAI)
Techniques enabling understanding of why an AI model made a given decision — critical for trust, auditing, and AI Act compliance.
Read more →F
Feature Engineering
Feature engineering transforms raw data into meaningful input variables that improve AI model performance and predictive accuracy.
Read more →Federated Learning
A distributed training approach that enables AI models to learn from decentralized data without sharing raw data between parties.
Read more →Fine-tuning
Retraining an AI model on specialized data — adapting a general foundation model to a specific domain or task.
Read more →Foundation Model
Large, pre-trained AI model serving as the foundation — customized via fine-tuning for specific applications.
Read more →Function Calling
An LLM capability that enables models to invoke external tools and APIs by generating structured function calls.
Read more →G
Generative AI
AI systems capable of creating new content including text, images, code, audio, and video from learned patterns.
Read more →GPU and TPU for AI
Specialized processors that accelerate AI model training and inference through massive parallel computation capabilities.
Read more →Grounding AI
Technique of anchoring AI model responses in factual data — eliminating hallucinations by providing context from reliable sources.
Read more →H
Human-in-the-Loop
Design pattern where a human verifies and approves AI decisions — quality control and safety.
Read more →Hyperautomation
An enterprise strategy combining AI, RPA, and multiple automation technologies to automate as many business processes as possible.
Read more →I
Information Retrieval for AI
The science and practice of finding relevant information from large collections to provide AI systems with accurate, grounded knowledge.
Read more →Intelligent Document Processing (IDP)
AI-powered systems that automatically extract, classify, and process information from unstructured documents at scale.
Read more →K
Knowledge Distillation
A training technique where a smaller 'student' model learns to replicate the behavior of a larger 'teacher' model.
Read more →Knowledge Graph
A structured representation of entities and their relationships that enables AI systems to reason about connected information.
Read more →M
Machine Translation with AI
AI-powered translation systems that convert text between languages with increasing accuracy, fluency, and domain awareness.
Read more →MCP (Model Context Protocol)
Open standard for communication between AI models and external data sources and tools — the "USB-C for artificial intelligence."
Read more →Mixture of Experts (MoE)
An architecture where multiple specialized sub-networks handle different inputs, activating only relevant experts per query.
Read more →MLOps
MLOps combines machine learning and DevOps practices to automate and streamline the deployment, monitoring, and management of AI models in production.
Read more →Model Card
A model card is a standardized documentation framework that describes an AI model's capabilities, limitations, intended use, and evaluation results.
Read more →Model Poisoning
Model poisoning attacks compromise AI systems by manipulating the model's parameters or training process to introduce hidden malicious behaviors.
Read more →Model Quantization
A technique for reducing AI model size and computational requirements by using lower-precision numerical representations.
Read more →Model Registry
A model registry is a centralized repository for versioning, storing, and managing machine learning models throughout their lifecycle.
Read more →Model Serving
The infrastructure and practices for deploying trained AI models to production environments where they handle real-time requests.
Read more →Model Versioning
Model versioning tracks changes to AI models, their training data, and configurations to ensure reproducibility and enable reliable rollback.
Read more →Multi-Agent Systems
AI architecture where dozens of specialized agents collaborate on tasks — each with unique competencies and roles.
Read more →Multimodal AI
AI models processing text, images, audio, and video simultaneously — understanding context from multiple information sources.
Read more →Multimodal RAG
Retrieval-Augmented Generation that works across text, images, tables, and other data types for richer, more complete AI responses.
Read more →N
Neural Scaling Laws
Empirical relationships showing how AI model performance improves predictably with increases in model size, data, and compute.
Read more →NIS2 and Artificial Intelligence
NIS2 Directive in the AI context — cybersecurity requirements for companies using AI systems in critical infrastructure.
Read more →NLP (Natural Language Processing)
The AI discipline focused on enabling machines to understand, interpret, generate, and meaningfully interact with human language.
Read more →P
Predictive Maintenance
AI-driven approach to equipment maintenance that predicts failures before they occur, reducing downtime and costs.
Read more →Prompt Engineering
The practice of designing and optimizing input instructions to elicit accurate, relevant, and useful responses from AI models.
Read more →Prompt Injection
Attack injecting malicious instructions into AI model input data — to take control of its behavior.
Read more →R
RAG (Retrieval-Augmented Generation)
Technique combining information retrieval with generation — the AI answers based on current documents, not just its "memory."
Read more →Recommendation Systems
AI systems that predict and suggest relevant items, content, or actions based on user behavior and preferences.
Read more →Reranking
A second-stage retrieval process that reorders search results using a more sophisticated model to improve relevance and accuracy.
Read more →Responsible AI
Responsible AI is an organizational approach ensuring AI systems are developed and deployed ethically, fairly, transparently, and with accountability.
Read more →RLHF (Reinforcement Learning from Human Feedback)
A training methodology that aligns AI models with human preferences using feedback-driven reinforcement learning.
Read more →ROI from AI
Frameworks and methods for measuring the return on investment from AI projects, including both quantitative and qualitative benefits.
Read more →RPA vs AI
Comparing Robotic Process Automation with AI-driven automation — their strengths, limitations, and how they complement each other.
Read more →S
Scaling AI in Organizations
Moving AI from isolated pilots to enterprise-wide adoption, addressing technical, organizational, and cultural challenges at scale.
Read more →Semantic Caching
An intelligent caching strategy that stores and retrieves AI responses based on meaning similarity rather than exact query matches.
Read more →Semantic Search
Search technology that understands the meaning and intent behind queries rather than just matching keywords.
Read more →Sentiment Analysis
AI technology that automatically detects and classifies emotional tone and opinions in text data at scale.
Read more →Shadow AI
Unauthorized use of AI tools by employees — without IT department knowledge or control, risking data leaks.
Read more →SLM (Small Language Models)
Compact AI models (1-7B parameters) running locally, fast, and cheaply — ideal for specialized tasks without cloud costs.
Read more →Speech-to-Text and Text-to-Speech
AI technologies that convert spoken language to written text and vice versa, enabling voice interfaces and accessibility solutions.
Read more →Structured Output
Techniques for constraining AI model responses to follow specific formats like JSON, XML, or predefined schemas.
Read more →Synthetic Data
Artificially generated datasets preserving statistical properties of originals — for AI training without privacy violations.
Read more →T
Temperature and Top-P Sampling
Key parameters that control the randomness and creativity of AI model text generation outputs.
Read more →Total Cost of Ownership for AI
Understanding the full financial picture of AI initiatives, from infrastructure and licensing to maintenance and talent.
Read more →Transfer Learning
A machine learning approach where knowledge gained from one task is applied to improve performance on a different but related task.
Read more →Transformer Architecture
The foundational neural network architecture behind modern large language models, based on self-attention mechanisms.
Read more →V
Vector Database
Specialized database storing data as numerical vectors — enabling semantic search for "similar" content.
Read more →Vendor Lock-In in AI
The risk of becoming overly dependent on a single AI vendor's proprietary technology, making switching costly or impractical.
Read more →Vibe Coding
Creating software by describing in natural language — the developer says "what," AI generates "how."
Read more →