Enterprise AI Glossary
The definitive dictionary of Artificial Intelligence concepts, technologies, and strategies, decoded for business leaders.
Showing 83 of 83 results
A
Agentic AI
Agentic AI refers to artificial intelligence systems designed to act autonomously, making decisions and executing multi-step workflows without constant human intervention.
AGI (Artificial General Intelligence)
AGI (Artificial General Intelligence) refers to a hypothetical AI system capable of understanding, learning, and applying intelligence across all cognitive tasks at a human or superhuman level.
Alignment
Alignment is the research and engineering process of ensuring an AI system's goals, behaviors, and outputs match human values and intentions.
Attention Mechanism
The attention mechanism is a neural network component that allows models to dynamically weigh the importance of different parts of an input sequence when generating an output.
B
Backpropagation
Backpropagation is the core algorithm for training neural networks, computing the gradient of the loss function with respect to the network's weights.
Bias in AI
AI bias refers to systematic errors in machine learning outputs that result from prejudiced assumptions or imbalanced data during training.
C
Chatbot
A chatbot is a software application designed to simulate human conversation, often utilizing large language models to provide dynamic, context-aware responses.
Chinchilla Scaling Laws
Chinchilla Scaling Laws state the optimal ratio of model parameters to training data, emphasizing that many early large models were undertrained given their size.
Computer Vision
Computer Vision is a subfield of artificial intelligence focused on enabling machines to extract meaningful information from digital images, videos, and other visual inputs.
D
Data Governance
Data Governance encompasses the policies and frameworks used to manage data availability, usability, integrity, and security across an organization.
Deep Learning
Deep Learning is a specialized branch of machine learning utilizing multi-layered artificial neural networks to automatically extract features from complex data.
Diffusion Models
Generative AI models that learn to synthesize data by iteratively reversing a gradual noise-adding process.
E
Edge AI
The deployment of AI algorithms locally on hardware devices rather than on centralized cloud infrastructure.
Embeddings
Dense numerical vectors that represent the semantic meaning of data, allowing AI models to measure relatedness.
Epoch
One complete pass of the entire training dataset through a machine learning algorithm during the training process.
Evaluating AI
The systematic assessment of an artificial intelligence model's performance, safety, robustness, and alignment before and after deployment.
F
Feature Engineering
The process of using domain knowledge to select, extract, and transform raw data into meaningful variables for machine learning models.
Feature Store
A centralized repository for storing, managing, and serving machine learning features for both training and inference.
Federated Learning
A decentralized machine learning approach where a model is trained across multiple devices or servers holding local data samples, without exchanging the raw data.
Few-Shot Learning
A technique where an AI model is guided to perform a specific task by providing it with a small number of examples, often directly in the prompt.
Fine-Tuning
The process of continuing the training of a pre-trained AI model on a specialized dataset to optimize it for targeted tasks or domains.
Foundation Model
A large-scale artificial intelligence model trained on a vast quantity of unlabelled data that serves as a versatile base for many downstream tasks.
G
Generative Adversarial Network (GAN)
A machine learning framework where two neural networks compete to generate synthetic data.
Generative AI
Artificial intelligence systems capable of generating novel text, images, code, or other media based on learned patterns.
Gradient Descent
An optimization algorithm used to minimize the error of a machine learning model by iteratively adjusting its parameters.
Grounding
The practice of linking an AI model's outputs to verifiable external sources or databases to improve accuracy and reduce hallucination.
H
Hallucination
A phenomenon where an AI model generates false, nonsensical, or unverified information while presenting it as fact.
Human-in-the-Loop (HITL)
A system design where human oversight is integrated into an AI workflow to review, correct, or approve outputs.
Hyperautomation
A business-driven approach to rapidly identify, vet, and automate as many processes as possible using multiple technologies.
Hyperparameter
A configuration setting used to control the learning process of a machine learning model, set before training begins.
I
Inference
The process of running live data through a trained machine learning model to make a prediction or generate an output.
Instruction Tuning
A fine-tuning process that trains a base language model to follow natural language commands and instructions.
Intent Recognition
The process of determining a user's goal or purpose from natural language input.
K
L
LangChain
An open-source orchestration framework designed to simplify the development of applications powered by language models.
Llama
A family of open-weight large language models developed and released by Meta.
LLM (Large Language Model)
A deep learning model trained on vast amounts of text data, designed to understand, generate, and interact with natural language.
Loss Function
A mathematical function that quantifies the difference between an AI model's predicted output and the actual true values.
M
Machine Learning (ML)
A field of artificial intelligence that uses algorithms and statistical models to enable computer systems to learn from data.
MCP (Model Context Protocol)
An open standard designed to connect AI models with external data sources and tools uniformly.
Model Decay
The degradation of an AI model's predictive performance over time as real-world data distributions change from the original training data.
Model Ops (MLOps)
A set of practices that combines machine learning, DevOps, and data engineering to reliably deploy and maintain ML models in production.
Model Registration
The process of cataloging and versioning AI models within an MLOps pipeline to track metadata, lineage, and deployment status.
Multimodal AI
Artificial intelligence systems capable of processing, understanding, and generating multiple types of data—such as text, images, audio, and video—simultaneously.
N
Named Entity Recognition (NER)
A natural language processing (NLP) technique that identifies and classifies key entities in unstructured text into predefined categories.
Neural Network
A computing system inspired by the biological neural networks that constitute animal brains, designed to recognize patterns and solve complex problems.
NLP (Natural Language Processing)
The branch of artificial intelligence concerned with giving computers the ability to understand, interpret, and generate human language.
O
Open Source AI
AI models, tools, and frameworks whose source code and weights are made publicly available for use, study, and modification.
OpenAI
A prominent AI research and deployment company known for creating the GPT series of large language models and other generative AI tools.
Orchestration
The automated configuration, management, and coordination of complex computer systems, AI models, and APIs to execute a cohesive workflow.
Overfitting
Overfitting occurs when an AI model learns the training data so well that it memorizes noise instead of general patterns, leading to poor performance on unseen data.
P
Parameter-Efficient Fine-Tuning (PEFT)
Parameter-Efficient Fine-Tuning (PEFT) adapts pre-trained language models for specific tasks by updating only a small subset of parameters, reducing computational and memory requirements.
Parameters
Parameters are the internal variables, including weights and biases, that a machine learning model learns from training data to make predictions.
Predictive Analytics
Predictive analytics leverages historical data, statistical algorithms, and machine learning techniques to estimate the likelihood of future outcomes.
Prompt Engineering
Prompt engineering is the systematic process of designing, refining, and optimizing text inputs to elicit accurate and relevant outputs from large language models.
Q
Q-Learning
Q-Learning is a model-free reinforcement learning algorithm used to learn the optimal action-selection policy for a given finite Markov decision process.
QLoRA (Quantized Low-Rank Adaptation)
QLoRA (Quantized Low-Rank Adaptation) is a parameter-efficient fine-tuning technique that backpropagates gradients through a quantized pre-trained model into low-rank adapters.
Quantization
Quantization is the process of mapping continuous infinite values to a smaller set of discrete finite values, used in AI to reduce the precision of model weights to decrease memory usage and improve inference speed.
R
RAG (Retrieval-Augmented Generation)
Retrieval-Augmented Generation (RAG) is an architectural framework that improves language model outputs by retrieving relevant factual information from an external database before generating a response.
Recurrent Neural Network (RNN)
A Recurrent Neural Network (RNN) is a class of artificial neural networks where connections between nodes can create a cycle, allowing output from some nodes to affect subsequent input to the same nodes.
Red Teaming
Actively testing an AI system by simulating adversarial attacks to discover vulnerabilities.
Reinforcement Learning (RL)
A machine learning technique where an agent learns to make decisions by performing actions and optimizing for a reward signal.
Return on AI (ROAI)
The financial and operational value generated by an AI initiative compared to its total lifecycle costs.
S
Semantic Search
A search technique that matches queries based on contextual meaning and intent rather than exact keyword overlap.
Synthetic Data
Artificially generated data that mimics the statistical properties of real-world data, used to safely train AI models.
System Prompt
The foundational instructions given to an AI model defining its persona, constraints, and overarching goals.
T
Temperature
A configuration parameter that controls the randomness and creativity of an AI model's text generation.
Tokenization
The process of breaking text into smaller, computationally manageable units called tokens for an AI model to process.
Training Data
The dataset used to initially teach an AI model how to recognize patterns, make predictions, or generate text.
Transfer Learning
A machine learning technique where an AI model trained on one task is repurposed as the starting point for a different, related task.
Transformer Architecture
The foundational neural network architecture of modern LLMs that processes data sequences in parallel using self-attention mechanisms.
V
Validation Set
A subset of data held out during training to tune model hyperparameters and evaluate performance, helping prevent overfitting.
Vector Database
A database optimized for storing and querying high-dimensional vectors, enabling rapid semantic similarity searches.
Vision-Language Model (VLM)
An AI model capable of simultaneously processing and understanding both visual and textual inputs.
W
Weight Decay
A regularization technique used during model training to penalize large weights, helping prevent overfitting.
Weights
The internal numeric values in a neural network that are optimized during training to map inputs to accurate predictions.
Word2Vec
An early foundational technique for natural language processing that translates words into continuous mathematical vectors to capture semantic relationships.