Agent
A software system that can select and perform a sequence of actions toward a goal within defined tools and permissions.
Glossary
Plain-English AI terms
Definitions focus on durable concepts rather than current product names.
Algorithm
A defined procedure or method used to process information or solve a task.
Artificial general intelligence
A hypothetical system with broadly human-level capability across many domains; it has not been demonstrated.
Artificial intelligence
A broad label for computer systems that perform tasks involving prediction, perception, language, recommendation, generation, or decision support.
Attention
A transformer mechanism that weighs relationships among tokens or other elements of an input.
Bias
A systematic pattern of error or disadvantage that can arise from data, labels, objectives, design, or use.
Chatbot
A conversational interface that may use rules, retrieval, language models, or a combination.
Computer vision
Methods that analyze images or video to recognize, classify, locate, or generate visual content.
Context window
The amount of material a model can consider during one request.
Deep learning
Machine learning using neural networks with many processing layers.
Embedding
A numerical vector representing learned relationships among words, images, records, or other items.
Explainability
Methods and information intended to help people understand why a model or system produced an output.
Fine-tuning
Additional training that adjusts a model using a curated dataset for a narrower behaviour or task.
Foundation model
A large model trained broadly and adapted to many downstream tasks.
Generative AI
AI that creates new text, images, audio, video, code, or other content.
Hallucination
A plausible-looking but unsupported, inaccurate, or fabricated AI output.
Inference
Running a trained model on new input to produce an output.
Large language model
A model trained on extensive language data to predict and generate token sequences.
Machine learning
Methods that use data to estimate patterns rather than relying only on hand-written rules.
Model
A learned mathematical structure that transforms inputs into predictions, scores, or generated outputs.
Multimodal AI
A system that processes or generates more than one type of data, such as text, images, and audio.
Natural language processing
Methods for analyzing, interpreting, transforming, or generating human language.
Neural network
A layered mathematical model whose parameters are adjusted during training.
Parameter
A learned numerical value inside a model.
Prompt
Instructions and context supplied to a generative AI system.
Reinforcement learning
Learning through actions, environmental feedback, and rewards or penalties.
Retrieval-augmented generation
A design that retrieves relevant source material and supplies it to a generative model at request time.
Supervised learning
Training with examples paired with expected labels or outputs.
Token
A unit of text or other data processed by a model.
Training data
Examples used to adjust a model’s parameters.
Transformer
A neural-network architecture built around attention and widely used for language and multimodal models.
Transparency
Disclosure and documentation about purpose, data, limitations, ownership, and use of an AI system.