AI Glossary
50 AI Terms, Explained in Plain English
A no-jargon glossary written for business leaders — not engineers. Every term is paired with a real-world example and a note on why it matters commercially.
Basics
The Foundations
Core concepts every business leader should understand.
Algorithm
An algorithm is a precise, step-by-step procedure for solving a problem or performing a computation. In the context of AI, algorithms define how a model learns from data, how it makes predictions, and how it optimises its performance.
Artificial Intelligence
Artificial Intelligence (AI) is a branch of computer science focused on building systems that perform tasks which normally require human intelligence, such as understanding language, recognising patterns, making decisions, and learning from experience over time.
Deep Learning
Deep Learning is a branch of machine learning that uses multi-layered neural networks to learn highly complex patterns directly from raw data such as images, audio and text, without the need for hand-crafted feature engineering.
Inference
Inference is the phase in which a trained model is used to produce predictions or outputs on new data. While training happens once (or periodically), inference happens every time a user interacts with an AI system, making it the dominant cost in most production deployments.
Machine Learning
Machine Learning (ML) is a subfield of AI in which systems learn patterns from historical data rather than following explicitly programmed rules, enabling them to make predictions or decisions on new, unseen data as conditions evolve.
Model
A model is the trained output of a machine learning process — a collection of learned parameters that, combined with an algorithm, can turn new inputs into predictions or generated content without being explicitly programmed for each case.
Neural Network
A neural network is a computational model loosely inspired by the human brain, consisting of interconnected layers of nodes (neurons) that transform inputs into outputs through weighted mathematical operations learned during training.
Supervised Learning
Supervised learning is a machine learning approach in which the model is trained on a dataset containing inputs paired with their correct outputs (labels). The model learns to map inputs to outputs, enabling it to predict labels for new, unseen examples.
Training Data
Training data is the dataset used to teach a machine learning model the patterns it needs to perform its task. The quality, quantity, diversity and recency of training data directly determine how accurate and fair the resulting model will be.
Unsupervised Learning
Unsupervised learning is a machine learning approach where the model learns patterns and structure from unlabelled data. Rather than predicting a known target, it uncovers groupings, anomalies or compressed representations hidden in the data.
Technical
How It Works
How modern AI systems are actually built and deployed.
Embedding
An embedding is a numerical vector representation of text, images or other data that captures semantic meaning. Items with similar meaning produce similar vectors, which makes embeddings the backbone of semantic search, recommendations and RAG systems.
Fine-tuning
Fine-tuning is the process of further training a pre-trained model on a smaller, task-specific dataset so that it specialises in a particular domain, tone or set of behaviours. It sits between using a base model as-is and training a new model from scratch.
Hallucination
A hallucination is when an AI model produces output that sounds plausible but is factually incorrect, fabricated or inconsistent with its sources. Hallucinations are a fundamental property of current generative models and the single biggest risk in enterprise deployments.
Large Language Model (LLM)
A Large Language Model (LLM) is a type of neural network trained on vast quantities of text to understand and generate human language. LLMs power chatbots, copilots, content generators and many modern AI features across consumer and business software.
Prompt Engineering
Prompt engineering is the practice of designing the text instructions given to a language model to produce reliable, accurate and appropriate outputs. Good prompts unlock significantly better performance without any change to the underlying model.
RAG (Retrieval Augmented Generation)
Retrieval Augmented Generation (RAG) is an architecture that combines a language model with an external knowledge source. Before generating an answer, the system retrieves relevant documents and feeds them to the model as context, dramatically reducing hallucinations and keeping answers current.
Temperature
Temperature is a parameter that controls how random or deterministic a language model's output is. Lower temperatures produce focused, predictable outputs; higher temperatures encourage diversity and creativity but increase the risk of hallucination or off-topic drift.
Token
A token is the basic unit of text that a language model processes. Tokens are usually subword chunks — roughly four characters or three-quarters of a word in English — and both the size of the model's context window and its pricing are typically measured in tokens.
Transformer
The transformer is a neural network architecture introduced in 2017 that uses a mechanism called self-attention to process sequences in parallel. It is the foundational architecture behind nearly all modern large language models and many leading vision and audio models.
Vector Database
A vector database is a specialist datastore designed to index and query high-dimensional vectors, typically embeddings, at scale. It enables fast similarity search — retrieving the most semantically related items for a given query vector — across millions or billions of items.
Business
Strategy & Governance
Strategy, governance and commercial concepts.
AI Ethics
AI ethics is the discipline of ensuring AI systems are designed, deployed and monitored in ways that respect fairness, transparency, privacy, human autonomy and wider societal impact. It goes beyond legal minimums to reflect organisational values.
AI Governance
AI governance is the set of policies, roles, controls and oversight mechanisms that ensure AI is used responsibly, safely and in line with law and organisational values. Effective governance is proportionate — tight where risk is high, light where risk is low.
AI Policy
An AI policy is an internal document that sets out how employees and vendors may — and may not — use AI within an organisation. A strong policy covers approved tools, prohibited use cases, data handling, intellectual property, disclosure, and responsibilities.
AI Readiness
AI readiness is an organisation's current capability to adopt AI successfully. It covers data quality, infrastructure, skills, processes, leadership alignment and governance. Low readiness is the most common cause of disappointing AI outcomes.
AI ROI
AI ROI is the measurable financial return an organisation generates from its AI investment, typically expressed as net benefits divided by total cost of ownership over a defined period. Realistic ROI calculations include build, run, governance and change-management costs.
AI Strategy
An AI strategy is an organisation's plan for where, why and how it will use AI to create value. A good AI strategy links specific business outcomes to prioritised use cases, required capabilities, governance guardrails, and a realistic investment roadmap.
Automation
Automation is the use of technology to perform tasks that would otherwise require human effort. AI extends traditional automation by handling tasks that involve judgment, unstructured data or natural language — not just deterministic, rule-based steps.
Chatbot
A chatbot is a software application that engages in conversation with users via text or voice. Modern AI-powered chatbots can answer questions, perform actions, escalate complex issues, and learn from each interaction to improve over time.
Customer Intelligence
Customer intelligence is the practice of combining data from every customer touchpoint and applying analytics and AI to produce a clearer picture of who customers are, what they want, and how they are likely to behave next.
Virtual Assistant
A virtual assistant is an AI system that helps an individual or team complete tasks, usually across multiple tools. Unlike a single-purpose chatbot, a virtual assistant combines conversation, context and actions to work alongside its user throughout the day.
Advanced
Beyond the Basics
Specialised techniques and emerging capabilities.
Agentic AI
Agentic AI refers to systems that can pursue goals autonomously by planning, taking actions across tools, observing outcomes and adapting their approach. Agentic systems go beyond single-turn question answering to execute multi-step workflows on a user's behalf.
Classification
Classification is a supervised learning task where the model assigns inputs to one of a predefined set of categories. Binary classification distinguishes between two classes; multi-class and multi-label variants handle more complex labelling problems.
Clustering
Clustering is an unsupervised learning technique that groups similar items together without any predefined labels. It is useful for discovering structure — customer segments, usage patterns, anomaly groups — that humans have not yet categorised.
Computer Vision
Computer vision is the branch of AI concerned with enabling machines to interpret and act on visual information. It powers applications from quality inspection and medical imaging to retail analytics, autonomous vehicles and augmented reality.
Explainable AI
Explainable AI (XAI) is a set of techniques and design practices that make AI systems' decisions understandable to humans. Explainability supports trust, debugging, accountability and regulatory compliance, particularly in consequential decisions.
Federated Learning
Federated learning is an approach to training machine learning models across multiple decentralised devices or servers, without moving the underlying data to a central location. Only model updates are exchanged, preserving privacy and data sovereignty.
Few-shot Learning
Few-shot learning is when a model learns a new task from a small number of examples, often supplied directly in the prompt. It sits between zero-shot (no examples) and full fine-tuning (many examples) and offers an excellent balance of quality and effort.
Generative AI
Generative AI refers to models that produce new content — text, images, code, audio, video, 3D — rather than merely classifying or predicting. Generative AI has fundamentally reshaped knowledge work, creative production and software engineering since 2022.
Knowledge Graph
A knowledge graph is a structured representation of entities (people, products, places, concepts) and the relationships between them. Knowledge graphs give AI systems explicit, queryable context that complements the implicit knowledge inside language models.
MLOps
MLOps is the discipline of operationalising machine learning: the practices, tools and culture needed to deploy, monitor, retrain and govern models reliably in production. It extends DevOps thinking to the unique challenges of data and models.
Multimodal AI
Multimodal AI refers to models that can process and generate multiple data types — text, images, audio, video — within a single system. They unlock workflows that were previously stitched together across many separate models.
Natural Language Processing (NLP)
Natural Language Processing is the field of AI concerned with interpreting, understanding and generating human language. NLP underpins chatbots, translation, summarisation, sentiment analysis, voice assistants and much of the productivity software UK teams now rely on daily.
OCR (Optical Character Recognition)
OCR is the technology that extracts machine-readable text from images and scanned documents. Modern OCR is now closely paired with layout understanding and AI extraction, turning scanned documents into structured data for downstream systems.
Prompt Injection
Prompt injection is a class of security vulnerability where a malicious input manipulates an LLM into ignoring its instructions, leaking confidential information, or taking unintended actions. It is the most important emerging AI security risk for organisations.
Regression
Regression is a supervised learning task where the model predicts a continuous numeric value rather than a category. Typical applications include forecasting demand, estimating prices, predicting customer lifetime value and modelling equipment wear.
Reinforcement Learning
Reinforcement Learning (RL) is a machine learning paradigm in which an agent learns to make decisions by interacting with an environment and receiving rewards or penalties. Over time, the agent learns a policy that maximises long-term reward.
Sentiment Analysis
Sentiment analysis uses NLP techniques to identify the emotional tone of text — positive, negative or neutral, and often more nuanced categories such as frustration, enthusiasm or sarcasm. It turns unstructured opinion into quantitative signal.
Speech Recognition
Speech recognition, also known as automatic speech recognition (ASR), converts spoken audio into text. Modern ASR handles accents, background noise, multiple speakers and domain-specific vocabulary far better than systems from even a few years ago.
Transfer Learning
Transfer learning is the practice of taking a model trained for one task and adapting it to a related task, saving time, data and compute. Fine-tuning a pre-trained language model on your own data is the most common contemporary example.
Zero-shot Learning
Zero-shot learning is when a model performs a task it was never explicitly trained on, by generalising from related knowledge. Modern LLMs routinely perform useful zero-shot classification, extraction and generation tasks with no task-specific training examples.
Ready to turn theory into results?
WayaNerd helps UK organisations move from curiosity to measurable ROI. Book a free strategy call and we will map the shortest route from where you are to your first production-ready AI use case.
Book a Free Strategy Call