Look it up
AI Glossary
AI terms. Simplified.
Browse by
A
- A/B test
- Ablation study
- Abstractive summarization
- Abuse detection
- Accelerator
- Accountability
- Accuracy
- Action
- Action item extraction
- Action space
- Activation function
- Active learning
- Actor-critic
- Adaptive learning
- Advanced driver-assistance system
- Adversarial example
- Adversarial machine learning
- Adversarial patch
- Adversarial testing
- Agent assist
- Agent evaluation
- Agent handoff
- Agent harness
- Agent identity
- Agent loop
- Agent marketplace
- Agent skill
- Agent sprawl
- Agent state
- Agent washing
- Agent-to-Agent protocol
- Agentic AI
- Agentic AI risk
- Agentic commerce
- Agentic workflow
- AI adoption
- AI agent
- AI anthropomorphism
- AI anxiety
- AI arms race
- AI assistant
- AI augmentation
- AI automation
- AI avatar
- AI backlash
- AI benefits eligibility screening
- AI browser
- AI bubble
- AI capability threshold
- AI carbon footprint
- AI champion
- AI coding assistant
- AI companion
- AI compliance
- AI content detection
- AI control
- AI crawler
- AI data center
- AI decision support
- AI demand forecasting
- AI development moratorium
- AI disclosure
- AI divide
- AI doomer
- AI energy consumption
- AI existential risk
- AI fluency
- AI fraud detection
- AI generation artifact
- AI glasses
- AI governance
- AI humanizer
- AI image upscaling
- AI incident
- AI incident reporting
- AI inventory
- AI job displacement
- AI literacy
- AI literacy obligation
- AI maturity
- AI meeting assistant
- AI Mode
- AI model provider
- AI model risk
- AI operating model
- AI Overviews
- AI pair programming
- AI PC
- AI personality
- AI personalization
- AI policy
- AI preparedness framework
- AI procurement
- AI productivity monitoring
- AI productivity paradox
- AI protein structure prediction
- AI readiness
- AI regulation
- AI regulatory sandbox
- AI resume screening
- AI roadmap
- AI safety
- AI safety case
- AI search visibility
- AI security
- AI security posture management
- AI sentience
- AI situational awareness
- AI slop
- AI strategy
- AI summary drift
- AI trainer
- AI training data transparency
- AI training opt-out
- AI transformation
- AI TRiSM
- AI use case
- AI verification burden
- AI water footprint
- AI winter
- AI-enabled phishing
- AI-native
- AIOps
- Aleatoric uncertainty
- Algorithm
- Algorithmic impact assessment
- Algorithmic literacy
- Algorithmic management
- Algorithmic pricing
- Algorithmic trading
- Algorithmic wage discrimination
- Alignment
- Alignment tax
- Ambient AI
- Ambient clinical documentation
- Anomaly detection
- Anonymization
- Answer engine
- Answer engine optimization
- API
- API gateway
- API key
- Approval gate
- Architecture
- Artificial general intelligence
- Artificial intelligence
- Artificial superintelligence
- Attack surface
- Attention
- Audio classification
- Audio model
- Audio source separation
- Audio-to-audio
- Audit trail
- Auditability
- Authentication
- Authentication token
- Authorization
- Autocomplete
- Autoencoder
- Automated content tagging
- Automated decision-making
- Automated machine learning
- Automated video interview
- Automatic speech recognition
- Automation
- Automation bias
- Automation paradox
- Autonomous agent
- Autonomous vehicle
- Autonomy level
- Autoregressive model
- Autoscaling
B
- Backdoor
- Backpropagation
- Bag of words
- Bagging
- Balanced accuracy
- Baseline model
- Batch
- Beam search
- Bearer token
- Benchmark
- Benchmark saturation
- Benchmarketing
- BERTScore
- Bi-encoder
- Bias
- Bias term
- Bias-variance tradeoff
- Binary classification
- Black-box model
- BLEU
- BM25
- Boosting
- Bot account
- Bounding box
- Branch
- Branching logic
- Brier score
- Bring your own AI
- Browser agent
- Build
- Business process automation
C
- Caching
- Calibration curve
- Capability
- Capability discovery
- Capability overhang
- Catastrophic AI risk
- Catastrophic forgetting
- Categorical feature
- Causal AI
- Causal inference
- Centaur workflow
- Center of excellence
- Chain of thought
- Chain-of-thought faithfulness
- Chain-of-thought prompting
- Change management
- Character card
- Chat history
- Chat model
- Chatbot
- Chatbot containment rate
- Checkpoint
- Chief AI officer
- Chinese room
- Chunk overlap
- Chunk size
- Chunking
- Churn prediction
- CI/CD
- Citation grounding
- Citation hallucination
- Citation rate
- Class imbalance
- Classification (machine learning)
- Classification threshold
- Classifier-free guidance
- Client library
- Clinical decision support
- Closed model
- Cloud computing
- Cloud provider
- Cluster
- Clustering
- Code completion
- Code generation
- Code review
- Cognitive offloading to AI
- Cold start
- Collaborative filtering
- Commit
- Common Crawl
- Completion
- Compute
- Compute efficiency
- Compute governance
- Computer use
- Computer vision
- Concept drift
- Concurrency
- Condition
- Conformity assessment
- Confounding
- Confusion matrix
- Connector
- Consent
- Constitutional AI
- Container
- Content moderation
- Content provenance
- Content-based filtering
- Context compaction
- Context engineering
- Context provider
- Context rot
- Context source
- Context window
- Contextual retrieval
- Continual learning
- Contract review AI
- Contrastive learning
- Conversation intelligence
- Conversation turn
- Conversational AI
- Convolution
- Convolutional neural network
- Copilot pattern
- Coreference resolution
- Cosine similarity
- Cost per million tokens
- Cost per request
- Counterfactual explanation
- CPU
- Credit scoring AI
- Critique prompt
- Cron job
- Cross-encoder
- Cross-modal retrieval
- Cross-validation
- Curriculum learning
- Custom assistant
- Custom instructions
D
- Dartmouth workshop
- Data annotation
- Data augmentation
- Data catalog
- Data cleaning
- Data dignity
- Data drift
- Data exfiltration
- Data governance
- Data ingestion
- Data labeling
- Data lake
- Data lakehouse
- Data leakage
- Data lineage
- Data minimization
- Data pipeline
- Data poisoning
- Data privacy
- Data processing agreement
- Data protection impact assessment
- Data provenance
- Data quality
- Data residency
- Data transformation
- Data warehouse
- Database
- Dataset
- Dataset card
- DBSCAN
- Dead internet theory
- Debugging
- Deceptive alignment
- Decision boundary
- Decision node
- Decision tree
- Decode phase
- Decoder (neural network)
- Decomposition
- Deep learning
- Deep neural network
- Deep reinforcement learning
- Deep research
- Deepfake
- Delimiters
- Denoising
- Dense retrieval
- Dependency
- Dependency parsing
- Deployment
- Depth estimation
- Deskilling
- Deterministic output
- Differential privacy
- Diffusion model
- Diffusion steps
- Diffusion transformer
- Digital twin
- Digital worker
- Dimensionality reduction
- Direct preference optimization
- Discriminative model
- Disparate impact
- Distillation
- Distributed training
- Document database
- Document question answering
- Documentation requirement
- Domain adaptation
- Dropout
- Drug discovery AI
- Dual-use
E
- Early stopping
- Edge AI
- Effective accelerationism
- Eliza effect
- ELT
- Embedding
- Embedding model
- Embodied AI
- Emergence
- Emergent capability
- Encoder (neural network)
- Encryption
- Endpoint
- Engagement optimization
- Ensemble learning
- Enterprise AI
- Enterprise search
- Environment variable
- Epistemic uncertainty
- Epoch
- ETL
- EU AI Act
- Eval set
- Evaluation
- Evaluation data contamination
- Evasion attack
- Event-driven workflow
- Exact match
- Exception handling
- Executor
- Expert model
- Expert system
- Explainability
- Exploding gradient
- Exploration-exploitation tradeoff
- Exponential backoff
- Extended thinking
- Extractive summarization
F
- F1 score
- Face swap
- Face verification
- Facial recognition
- Fairness
- Faithfulness
- Fallback model
- False negative
- False positive
- Feature
- Feature engineering
- Feature importance
- Feature scaling
- Feature selection
- Federated learning
- Few-shot learning
- Few-shot prompting
- Filter bubble
- Fine-tuning
- Flow matching
- Foundation model
- Framework
- Frontier model
- Function calling
G
- General-purpose AI
- Generalization
- Generative adversarial network
- Generative AI
- Generative engine optimization
- Generative pre-trained transformer
- Ghost work
- Git
- GloVe
- Goal misgeneralization
- Golden dataset
- Governance committee
- GPAI Code of Practice
- GPAI model with systemic risk
- GPU
- Gradient
- Gradient boosting
- Gradient descent
- GraphRAG
- Greedy decoding
- Grid search
- Ground truth
- Grounding
- Grounding prompt
- Guardrail
H
I
- Idempotency
- Image captioning
- Image classification
- Image editing
- Image generation
- Image segmentation
- Image understanding
- Image-to-3D
- Image-to-image
- Impact assessment
- In-context learning
- Indirect prompt injection
- Inference
- Inference compute
- Inference cost
- Inference endpoint
- Inference-time scaling
- Information extraction
- Inpainting
- Input tokens
- Instance segmentation
- Instruction tuning
- Insurance underwriting AI
- Integration layer
- Integration test
- Intelligence explosion
- Intelligent automation
- Intelligent document processing
- Intelligent tutoring system
- Intent recognition
- Interoperability
- Interpretability
- Interpretable model
- ISO/IEC 42001
- Isolation
J
K
L
- L1 regularization
- L2 regularization
- Label
- Labeled data
- Large language model
- Latency
- Latency budget
- Latent diffusion
- Latent representation
- Latent semantic analysis
- Latent space
- Learning rate
- Learning to rank
- Least privilege
- Legal research AI
- Lemmatization
- Lethal trifecta
- Liar's dividend
- Linear regression
- Liveness detection
- LLM-as-a-judge
- Local tool
- Log loss
- Logistic regression
- Logit
- Long short-term memory
- Long-horizon task
- Long-running agent
- Long-term memory
- Loop
- LoRA
- Loss function
- Low-code
M
- Machine learning
- Machine translation
- Machine unlearning
- Manual review
- Masked language model
- Max tokens
- MCP client
- MCP prompt
- MCP resource
- MCP server
- MCP tool
- Mean absolute error
- Mean average precision
- Mean squared error
- Mechanistic interpretability
- Medical imaging AI
- Membership inference
- Memory
- Memory bandwidth
- Memory poisoning
- Merge
- Meta-learning
- Metadata filtering
- Middleware
- Missing value imputation
- Misuse
- Mixture of experts
- Modality
- Model
- Model alignment
- Model card
- Model collapse
- Model compression
- Model confidence score
- Model Context Protocol
- Model deprecation
- Model drift
- Model extraction
- Model governance
- Model inventory
- Model inversion
- Model leakage
- Model nondeterminism
- Model poisoning
- Model router
- Model serving
- Model version
- Model welfare
- Model-agnostic
- Monocular depth estimation
- Multi-agent system
- Multi-head attention
- Multi-turn conversation
- Multiclass classification
- Multilabel classification
- Multimodal AI
- Multimodal prompt
- Multitask learning
- Music generation
N
- N-gram
- Naive Bayes
- Named entity recognition
- Narrow AI
- Native multimodality
- Natural language generation
- Natural language processing
- Natural language understanding
- Negative instruction
- Neural network
- Neural processing unit
- Neuromorphic computing
- Next-token prediction
- NIST AI Risk Management Framework
- No-code
- Node
- NoSQL
- Numerical feature
O
P
- Package
- Panoptic segmentation
- Paperclip maximizer
- Parallel step
- Parallelism
- Parameter
- Parameter count
- Parameter-efficient fine-tuning
- Part-of-speech tagging
- Pass rate
- pass@k
- Pay-per-crawl
- Payload
- Perplexity
- Persona prompt
- PHI
- PII
- Pilot
- Planner
- Planning horizon
- Policy (reinforcement learning)
- Policy gradient
- Pooling layer
- Pose estimation
- Positional encoding
- Post-training
- Precision
- Precision agriculture
- Prediction calibration
- Prediction interval
- Predictive analytics
- Predictive lead scoring
- Predictive maintenance
- Predictive policing
- Preference data
- Prefill
- Pretraining
- Principal component analysis
- Private AI
- Production deployment
- Prohibited AI practice
- Prompt
- Prompt chaining
- Prompt engineering
- Prompt hygiene
- Prompt injection
- Prompt leakage
- Prompt library
- Prompt optimization
- Prompt sensitivity
- Prompt template
- Prompt tuning
- Prompt variable
- Prompt versioning
- Proof of concept
- Proof of personhood
- Propensity model
- Proprietary model
- Protocol
- Proximal policy optimization
- Proxy variable
- Pseudo-labeling
- Pull request
- Purpose limitation
Q
R
- Random forest
- Random search
- Random seed
- Rate limiting
- Re-ranking
- ReAct
- Real-time voice AI
- Reasoning effort
- Reasoning model
- Reasoning scratchpad
- Reasoning tokens
- Reasoning trace
- Recall
- Recommendation system
- Recordkeeping
- Recurrent neural network
- Red teaming
- Refactoring
- Reflection prompt
- Refusal
- Refusal rate
- Regression (machine learning)
- Regression test
- Regularization
- Reinforcement learning
- Reinforcement learning from AI feedback
- Reinforcement learning from human feedback
- Relation extraction
- Relational database
- Relevance
- ReLU
- Remote tool
- Repetition penalty
- Repository
- Representation learning
- Request
- Reskilling
- Resource URI
- Response
- Responsible AI
- REST API
- Retrieval
- Retrieval freshness
- Retrieval quality
- Retrieval-augmented generation
- Retriever
- Retry
- Return on investment
- Reward model
- Reward tampering
- Right to explanation
- Risk assessment
- Risk tier
- Robotic process automation
- Robotics foundation model
- Robustness
- ROC AUC
- ROC curve
- Role prompting
- Root mean squared error
- ROUGE
- Routing model
- Rubric
- Rule-based automation
S
- Safety filter
- Safety policy
- Sampling
- Sandbagging
- Sandbox
- Scalable oversight
- Scaling law
- Scheduled workflow
- Schema
- Scheming
- SDK
- Search over reasoning
- Secrets management
- Self-attention
- Self-consistency
- Self-hosted language model
- Self-supervised learning
- Self-training
- Semantic search
- Semantic segmentation
- Semi-structured data
- Semi-supervised learning
- Sentiment analysis
- Sequence-to-sequence model
- Server
- Serverless
- Service-level agreement
- Shadow AI
- Sharding
- Short-term memory
- Sigmoid function
- Similarity score
- Small language model
- Smart reply
- Softmax
- Sound event detection
- Sovereign AI
- Spam filter
- Sparse retrieval
- Sparsity
- Speaker attribution error
- Speaker diarization
- Speaker recognition
- Speaker verification
- Specification gaming
- Specificity
- Spectral clustering
- Speculative decoding
- Speech enhancement
- Speech recognition
- Speech synthesis
- Speech translation
- Speech-to-speech model
- Spiking neural network
- SQL
- Stacking
- State space model
- Stochastic parrot
- Stop sequence
- Stop words
- Streaming
- Structured data
- Structured output
- Structured prompt
- Student model
- Style transfer
- Subagent
- Supervised fine-tuning
- Supervised learning
- Supply-chain risk
- Support vector machine
- Surveillance pricing
- Sycophancy
- Symbolic AI
- Synthetic data
- Synthetic identity fraud
- Synthetic media
- Synthetic performer
- System card
- System prompt
T
- Table question answering
- Task queue
- Teacher model
- Technological singularity
- Temperature
- Temporary chat
- Tensor
- Test generation
- Test set
- Test-time compute
- Text analytics
- Text classification
- Text summarization
- Text-to-3D
- Text-to-audio
- Text-to-image
- Text-to-speech
- Text-to-video
- TF-IDF
- Theory of mind (AI)
- Third-party risk
- Threat model
- Throughput
- Time to first token
- Time-series forecasting
- Timeout
- TinyML
- Token
- Token cost
- Tokenization
- Tokenizer
- Tool approval
- Tool calling
- Tool error
- Tool permission
- Tool poisoning
- Tool registry
- Tool result
- Tool routing
- Tool schema
- Tool selection
- Tool use
- Top-k sampling
- Top-p
- Topic modeling
- TOPS
- Total cost of ownership
- TPU
- Train-test split
- Training
- Training compute
- Training data
- Training data consent
- Training data copyright
- Training data leakage
- Training data wall
- Training run
- Transfer learning
- Transformer
- Transparency
- Tree of thoughts
- Trigger
- True negative
- True positive
- Turing test
U
V
- Validation set
- Value function
- Vanishing gradient
- Vector database
- Vector search
- Vendor risk
- Version control
- Vertical AI agent
- Vibe coding
- Video generation
- Video understanding
- Vision model
- Vision transformer
- Vision-language model
- Visual question answering
- Voice activity detection
- Voice agent
- Voice assistant
- Voice barge-in
- Voice cloning
- Voice cloning fraud
- Voice deepfake
- Voting classifier
- VRAM