AGI (Artificial General Intelligence)
A hypothetical AI system that can perform most intellectual tasks at a human level across many domains.
AI reference term
The AI Glossary
Essential technical terminology explained at the highest level of clarity. Designed for researchers, students, and human-centered education.
Showing 213 matching terms.
AI Agent
A software system that can observe, reason, and take actions to achieve a goal, often using tools and memory.
AI reference term
AI Alignment
The work of making AI systems behave according to human intentions, values, and safety constraints.
AI reference term
AI Governance
Policies, standards, and oversight mechanisms that guide how AI is developed and used in society.
AI reference term
AI Safety
A field focused on reducing harmful behavior, failures, and misuse risks in AI systems.
AI reference term
Algorithm
A defined set of rules or steps that a computer follows to solve a problem or complete a task.
AI reference term
Algorithmic Bias
Systematic unfairness in model outputs caused by skewed data, assumptions, or modeling choices.
AI reference term
Algorithmic Transparency
How clearly an AI system's logic, data sources, and limitations are documented and understandable.
AI reference term
Annotation
Human-added labels or metadata used to train or evaluate machine learning models.
AI reference term
API (Application Programming Interface)
A structured way for one software system to send requests to and receive responses from another system.
AI reference term
Artificial Intelligence (AI)
The broad field of building systems that perform tasks requiring pattern recognition, reasoning, language, or decision-making.
AI reference term
Attention Mechanism
A model component that dynamically focuses on relevant parts of an input when producing an output.
AI reference term
Autonomous System
A system that can make decisions and act with limited or no direct human control in real time.
AI reference term
Backpropagation
The core training algorithm that updates model weights by propagating prediction errors backward through the network.
AI reference term
Baseline Model
A simple reference model used to compare whether more complex approaches actually improve results.
AI reference term
Benchmark
A standardized test or dataset used to measure and compare model performance.
AI reference term
Bias
A consistent pattern of error or unfairness in data or model behavior.
AI reference term
Big Data
Very large and complex datasets that require scalable storage and processing techniques.
AI reference term
Black-Box Model
A model whose internal reasoning is difficult to interpret directly by humans.
AI reference term
Calibration
How well a model's confidence scores match actual correctness probabilities.
AI reference term
Chain-of-Thought
A reasoning style where an AI model decomposes a problem into intermediate steps.
AI reference term
Classification
A task where a model assigns an input to one or more predefined categories.
AI reference term
Classifier
A model designed specifically for classification tasks.
AI reference term
CLIP
A multimodal model architecture that learns shared representations between text and images.
AI reference term
Compute
The processing resources required to train and run models, often measured in FLOPS or GPU hours.
AI reference term
Computer Vision
The branch of AI that extracts meaning from images and video.
AI reference term
Context Window
The maximum amount of input tokens a language model can process at once.
AI reference term
Continual Learning
Training approaches that let a model keep learning from new data without forgetting prior knowledge.
AI reference term
Convolutional Neural Network (CNN)
A neural architecture optimized for processing grid-like data such as images.
AI reference term
Cross-Entropy Loss
A common objective function used to train classification models by penalizing incorrect probabilities.
AI reference term
Data Augmentation
Techniques that create modified training examples to improve model generalization.
AI reference term
Data Drift
A shift in real-world input data over time that can degrade model performance.
AI reference term
Data Labeling
The process of assigning tags or target outputs to raw data for supervised learning.
AI reference term
Dataset
A collection of structured or unstructured examples used for training, validation, or testing.
AI reference term
Decision Boundary
The surface in feature space that separates classes predicted by a classifier.
AI reference term
Decision Tree
A model that makes predictions through a sequence of if-then feature splits.
AI reference term
Deep Learning
A subset of machine learning that uses many-layer neural networks for representation learning.
AI reference term
Diffusion Model
A generative architecture that learns to reverse noise to synthesize images, audio, or other content.
AI reference term
Distillation
Compressing knowledge from a large teacher model into a smaller student model.
AI reference term
Domain Adaptation
Methods that transfer a model trained in one domain to perform better in another domain.
AI reference term
Embedding
A numeric vector representation that captures semantic meaning of text, images, or other data.
AI reference term
Encoder
The component of a model that transforms input into latent representations.
AI reference term
Ensemble
Combining predictions from multiple models to improve robustness or accuracy.
AI reference term
Evaluation Set
A held-out dataset used to measure model quality after training.
AI reference term
Explainability
The degree to which a model's behavior can be interpreted and explained to humans.
AI reference term
False Negative
An incorrect prediction where a model misses a true positive case.
AI reference term
False Positive
An incorrect prediction where a model incorrectly flags a negative case as positive.
AI reference term
Feature
An input variable used by a model to make predictions.
AI reference term
Feature Engineering
Designing or transforming input variables to make learning easier and more effective.
AI reference term
Feature Extraction
Converting raw data into informative features that a model can use.
AI reference term
Few-Shot Learning
Learning or adapting behavior from only a small number of examples.
AI reference term
Fine-Tuning
Continuing training on domain-specific data to adapt a pre-trained model to a specific task.
AI reference term
Foundation Model
A large pre-trained model that can be adapted to many downstream tasks.
AI reference term
Function Calling
A model capability to generate structured calls that trigger external tools or APIs.
AI reference term
GAN (Generative Adversarial Network)
A generative setup where a generator and discriminator train against each other.
AI reference term
Generalization
How well a model performs on new, unseen data outside the training set.
AI reference term
Generative AI
AI systems that produce new content such as text, images, audio, video, or code.
AI reference term
Gradient
A vector showing how much each parameter should change to reduce loss.
AI reference term
Gradient Descent
An optimization method that updates parameters in the direction that reduces error.
AI reference term
Ground Truth
Trusted reference labels used to train or evaluate model outputs.
AI reference term
Guardrails
Rules, checks, and controls that limit unsafe or undesired model behavior.
AI reference term
Hallucination
When a model generates fluent but false or unsupported information.
AI reference term
Human-in-the-Loop
A workflow where humans review, guide, or override AI outputs.
AI reference term
Hyperparameter
A configuration value set before training, such as learning rate, batch size, or depth.
AI reference term
In-Context Learning
A model's ability to follow patterns from examples provided directly in the prompt.
AI reference term
Inference
The runtime phase where a trained model generates predictions or outputs.
AI reference term
Inference-Time Compute
The amount of processing power consumed while producing each response.
AI reference term
Instruction Tuning
Fine-tuning a model on instruction-response pairs to improve task following.
AI reference term
Intent Classification
Predicting the user's purpose from a text query to route it correctly.
AI reference term
Jailbreak
A prompt technique intended to bypass a model's safety constraints.
AI reference term
Knowledge Cutoff
The latest point in time reflected in a model's training data.
AI reference term
Knowledge Distillation
Training a smaller model to imitate the outputs of a larger model.
AI reference term
Knowledge Graph
A graph structure of entities and relationships used for reasoning or retrieval.
AI reference term
Label Smoothing
A regularization method that softens hard labels to improve generalization.
AI reference term
Latency
The time between sending a request and receiving the model's output.
AI reference term
Large Language Model (LLM)
A language model trained on massive text corpora to generate and analyze text.
AI reference term
Learning Rate
A training hyperparameter controlling how much parameters change each update step.
AI reference term
LoRA (Low-Rank Adaptation)
A parameter-efficient fine-tuning method that adds low-rank adapter matrices.
AI reference term
Loss Function
A mathematical objective that quantifies prediction error during training.
AI reference term
Machine Learning (ML)
Methods that allow systems to learn patterns from data and improve over time.
AI reference term
Memory (Agent Memory)
Stored context an AI agent uses across steps or sessions to improve continuity.
AI reference term
Mixture of Experts (MoE)
An architecture with specialized subnetworks where only selected experts run per input.
AI reference term
Model Card
Documentation describing a model's intended use, metrics, limitations, and risks.
AI reference term
Model Drift
Performance degradation over time as real-world conditions diverge from training assumptions.
AI reference term
Model Quantization
Reducing numeric precision of model weights to decrease memory and inference cost.
AI reference term
Multimodal Model
A model that can process or generate multiple data types such as text, image, and audio.
AI reference term
Named Entity Recognition (NER)
An NLP task that identifies entities such as people, places, dates, or organizations.
AI reference term
Natural Language Processing (NLP)
The branch of AI focused on understanding and generating human language.
AI reference term
Neural Network
A layered computational model inspired by biological neurons and synapses.
AI reference term
Normalization
Transforming values to a consistent scale to improve optimization stability.
AI reference term
OCR (Optical Character Recognition)
Technology that converts text in images or scans into machine-readable text.
AI reference term
Open-Source Model
A model released with public weights or code for inspection, adaptation, and reuse.
AI reference term
Overfitting
When a model memorizes training data and performs poorly on unseen inputs.
AI reference term
Parameter
A learned weight inside a model that influences its outputs.
AI reference term
Parameter-Efficient Fine-Tuning (PEFT)
Methods that adapt models by training a small subset of added parameters.
AI reference term
Perplexity
A language-model metric measuring how surprised the model is by true next tokens.
AI reference term
Pipeline
An ordered workflow of preprocessing, model steps, and postprocessing stages.
AI reference term
Precision
The proportion of predicted positives that are actually correct.
AI reference term
Pretraining
Initial large-scale model training on broad data before downstream adaptation.
AI reference term
Prompt
The input instructions and context provided to a generative model.
AI reference term
Prompt Engineering
Designing prompts to improve output quality, reliability, and controllability.
AI reference term
Prompt Injection
An attack pattern where malicious instructions are inserted into model inputs or retrieved content.
AI reference term
Pruning
Removing less important model weights or neurons to reduce size and compute.
AI reference term
Quantization
Converting model weights to lower precision formats such as 8-bit or 4-bit.
AI reference term
RAG (Retrieval-Augmented Generation)
A method that retrieves external knowledge and feeds it into generation at inference time.
AI reference term
Recall
The proportion of actual positives that a model correctly identifies.
AI reference term
Recommendation System
A model pipeline that predicts user preferences for ranking content or products.
AI reference term
Red Teaming
Stress-testing an AI system with adversarial prompts to reveal failures and risks.
AI reference term
Reinforcement Learning
Training by reward signals where an agent learns actions that maximize long-term return.
AI reference term
Reinforcement Learning from Human Feedback (RLHF)
A training method that uses human preference signals to shape model behavior.
AI reference term
Retrieval
Finding relevant documents or records from a knowledge source for a query.
AI reference term
Reward Model
A model that scores outputs based on preference signals, often used in RLHF pipelines.
AI reference term
Robustness
A model's ability to maintain performance under noise, shifts, or adversarial inputs.
AI reference term
Safety Filter
A moderation layer that blocks or rewrites unsafe model inputs or outputs.
AI reference term
Scaling Law
An empirical relationship showing how performance improves with model size, data, or compute.
AI reference term
Semantic Search
Search that matches meaning rather than exact keyword overlap, often using embeddings.
AI reference term
Self-Supervised Learning
Learning representations from unlabeled data by predicting masked or transformed parts.
AI reference term
Sentiment Analysis
An NLP task that classifies emotional tone or opinion in text.
AI reference term
Small Language Model (SLM)
A compact language model optimized for lower latency, cost, or on-device usage.
AI reference term
Sparse Model
A model where many parameters are zero or inactive to reduce computation.
AI reference term
Supervised Learning
Training a model with labeled examples that map inputs to known outputs.
AI reference term
Synthetic Data
Artificially generated data used to augment, simulate, or protect sensitive training data.
AI reference term
System Prompt
A high-priority instruction that sets behavior, policy, and response style for a model.
AI reference term
Temperature
A sampling setting controlling randomness in generated outputs.
AI reference term
Token
A chunk of text processed by language models, such as a word piece or symbol.
AI reference term
Tokenization
The process of splitting text into tokens for model input.
AI reference term
Tool Use
A model's ability to call external tools such as search, calculators, or APIs.
AI reference term
Top-k Sampling
A decoding strategy that samples only from the k most likely next tokens.
AI reference term
Top-p (Nucleus) Sampling
A decoding strategy that samples from the smallest token set whose probabilities sum to p.
AI reference term
Transfer Learning
Applying knowledge learned in one task or domain to improve another task.
AI reference term
Transformer
A neural architecture that uses attention to model relationships across sequences in parallel.
AI reference term
Training Loss
The model error value computed during training and optimized downward over time.
AI reference term
Unsupervised Learning
Learning patterns from unlabeled data without explicit target outputs.
AI reference term
Validation Set
A dataset used during development to tune models and prevent overfitting.
AI reference term
Vector Database
A database optimized for storing and querying high-dimensional embedding vectors.
AI reference term
Vision-Language Model (VLM)
A multimodal model that jointly processes visual and textual information.
AI reference term
Weak Supervision
Using noisy, heuristic, or partial labels to train models when clean labels are scarce.
AI reference term
Weight
A learned numeric value that scales signals passing through a neural network.
AI reference term
Word Embedding
A dense vector representation of words capturing semantic relationships.
AI reference term
XAI (Explainable AI)
Techniques and practices for making AI predictions more transparent and understandable.
AI reference term
Zero-Shot Learning
Solving tasks without task-specific examples by relying on prior general knowledge.
AI reference term
Agentic Workflow
A multi-step process where an AI system plans, executes, checks results, and iterates toward a goal.
AI reference term
AI Act
The European Union's risk-based regulatory framework for AI systems and providers.
AI reference term
Alignment Tax
The extra cost in time, compute, or product velocity required to make systems safer and more controllable.
AI reference term
Benchmark Leakage
When benchmark test examples or close variants are present in training data, inflating reported performance.
AI reference term
Causal Inference
Methods for estimating cause-and-effect relationships rather than simple correlations.
AI reference term
Confidence Interval
A statistical range that likely contains the true value of a measured model metric.
AI reference term
Constitutional AI
A training and behavior-shaping approach where model outputs are guided by a fixed set of written principles.
AI reference term
Data Lineage
A record of where data came from, how it was transformed, and where it is used.
AI reference term
Data Provenance
The documented origin, ownership, and history of a dataset or model artifact.
AI reference term
Differential Privacy
A privacy technique that adds statistical noise so individual records cannot be reliably inferred from outputs.
AI reference term
Distilled Model
A smaller model trained to imitate a larger model's behavior while using less compute at inference.
AI reference term
Embedding Model
A model specialized for converting data into vectors used for semantic search, clustering, and retrieval.
AI reference term
Eval Harness
A repeatable evaluation framework that runs prompts, datasets, and scoring logic across model versions.
AI reference term
Feature Store
A managed system for storing and serving validated ML features consistently for training and inference.
AI reference term
Groundedness
The degree to which an AI response is supported by source data or retrieved evidence.
AI reference term
Guarded Decoding
A generation strategy that constrains output tokens to valid structures or policy-compliant choices.
AI reference term
Human Preference Model
A model trained on human rankings to predict which responses users are likely to prefer.
AI reference term
Inference Endpoint
A deployed API interface that receives model requests and returns predictions in production.
AI reference term
Knowledge Base
A curated collection of documents or records used for retrieval, support automation, or grounding responses.
AI reference term
Latent Space
A compressed representational space where similar concepts are positioned near each other as vectors.
AI reference term
Model Registry
A central catalog for versioning, approving, and tracking models across environments.
AI reference term
On-Device AI
AI inference performed locally on user hardware rather than in a remote cloud service.
AI reference term
Output Parser
Logic that validates and converts model output into strongly typed, machine-usable structures.
AI reference term
Prompt Template
A reusable prompt pattern with variables, formatting rules, and task-specific instructions.
AI reference term
Retrieval Precision
The proportion of retrieved items that are relevant to the user's query.
AI reference term
Safety Case
A structured argument, supported by evidence, that an AI system is safe for a defined context of use.
AI reference term
Shadow Deployment
Running a model in parallel with production traffic without affecting user-facing decisions.
AI reference term
Structured Output
Model output constrained to a defined schema such as JSON, tool arguments, or typed fields.
AI reference term
Test-Time Compute
Additional inference computation used during response generation to improve quality or reasoning.
AI reference term
Trust Calibration
Aligning user confidence in AI outputs with the system's actual reliability in each task.
AI reference term
Usage-Based Billing
Pricing where costs scale with API calls, tokens, inference time, or consumed compute.
AI reference term
Zero Data Retention
A policy where request/response payloads are not stored after processing beyond short-lived operational windows.
AI reference term
Speculative Decoding
An inference acceleration method where a small draft model proposes tokens that a larger model verifies in parallel.
AI reference term
KV Cache
Stored key and value tensors from previous tokens that let transformers generate new tokens without recomputing past attention.
AI reference term
MCP (Model Context Protocol)
An open protocol that lets AI applications connect to external tools, data sources, and context providers in a standard way.
AI reference term
Agentic Loop
An iterative cycle where an AI agent observes, plans, acts, and reflects until it completes a goal or hits a stop condition.
AI reference term
ReAct
A prompting pattern that interleaves reasoning steps with tool-use actions to solve tasks more reliably.
AI reference term
Tree of Thoughts
A reasoning approach where a model explores multiple branching solution paths and selects the most promising ones.
AI reference term
DPO (Direct Preference Optimization)
A training method that fine-tunes models directly on preference pairs without needing a separate reward model.
AI reference term
QLoRA
A fine-tuning technique that combines 4-bit weight quantization with LoRA adapters to reduce memory needs.
AI reference term
Flash Attention
An optimized attention algorithm that reduces memory use and speeds up transformer training and inference.
AI reference term
Multi-Head Attention
A transformer mechanism that runs several attention operations in parallel to capture different types of relationships.
AI reference term
Positional Encoding
Information added to token embeddings so transformers can distinguish sequence order.
AI reference term
RoPE (Rotary Positional Embeddings)
A positional encoding method that rotates query and key vectors to encode relative token positions.
AI reference term
ALiBi
A positional bias method that penalizes attention scores based on token distance, helping models extrapolate to longer contexts.
AI reference term
Sliding Window Attention
An attention pattern where each token attends only to a fixed-size window of nearby tokens to reduce compute.
AI reference term
BPE (Byte Pair Encoding)
A subword tokenization algorithm that merges the most frequent character pairs into reusable tokens.
AI reference term
SentencePiece
A language-agnostic tokenizer that learns subword units directly from raw text without pre-splitting on whitespace.
AI reference term
ANN (Approximate Nearest Neighbor)
Algorithms that find vectors close to a query without exhaustive comparison, trading exactness for speed.
AI reference term
HNSW
A graph-based index structure for fast approximate nearest-neighbor search over high-dimensional vectors.
AI reference term
Reranker
A model that reorders an initial set of retrieved results to put the most relevant items at the top.
AI reference term
Hybrid Search
A retrieval approach that combines keyword (lexical) search with vector (semantic) search for better recall and precision.
AI reference term
Cross-Encoder
A model that scores a query and document together in one pass for high-accuracy relevance judgments.
AI reference term
Bi-Encoder
A model that encodes queries and documents into separate vectors so they can be compared quickly at scale.
AI reference term
LLM-as-Judge
Using a language model to score or compare outputs from other models during evaluation.
AI reference term
Pass@k
A code-evaluation metric measuring the chance that at least one of k generated samples passes the tests.
AI reference term
MMLU
A benchmark testing language models across 57 academic and professional subjects using multiple-choice questions.
AI reference term
HumanEval
A benchmark of Python programming problems used to measure code-generation correctness via unit tests.
AI reference term
GSM8K
A benchmark of grade-school math word problems used to evaluate step-by-step reasoning in language models.
AI reference term
Factuality
How accurately a model's claims match verifiable real-world information.
AI reference term
Citations
References to source passages or documents included in a model's response to support its claims.
AI reference term
Watermarking
Embedding a detectable signal in AI-generated text or media so it can later be identified as machine-produced.
AI reference term
Mid-training
An intermediate training phase between pretraining and post-training, often used for capability or domain adjustments.
AI reference term
Post-training
Training steps applied after pretraining, such as instruction tuning, preference optimization, and safety tuning.
AI reference term
Self-Play
A training setup where a model improves by generating data through interactions or competitions with copies of itself.
AI reference term
RAG-Fusion
A retrieval method that generates multiple query variants, retrieves results for each, and fuses the rankings.
AI reference term
Multi-Query Retrieval
A retrieval technique that rewrites the user query into several variants to improve recall.
AI reference term
Parent-Document Retrieval
A retrieval pattern that searches small chunks but returns their larger parent documents for richer context.
AI reference term
Beam Search
A decoding algorithm that keeps the top several candidate sequences at each step to find higher-probability outputs.
AI reference term
Repetition Penalty
A decoding setting that lowers the probability of tokens the model has already produced to reduce loops.
AI reference term
Frequency Penalty
A decoding setting that reduces the probability of tokens proportionally to how often they've appeared so far.
AI reference term
Presence Penalty
A decoding setting that reduces the probability of tokens that have appeared at all, encouraging new topics.
AI reference term