Introduction
AI has quickly become a part of everyday conversation, but the terminology can be daunting. Whether you’re new to the field or just need a refresher, this glossary explains common AI terms in clear, conversational language. Bookmark it for quick reference, and feel free to share it with colleagues who are just starting their AI journey.
Large language model (LLM)
A type of AI trained on massive amounts of text to understand and generate human‑like language. Popular LLMs include OpenAI’s GPT series and Google’s Gemini. They can answer questions, write content and translate languages.
Retrieval‑augmented generation (RAG)
A technique that combines an LLM with a search component. Instead of relying solely on the model’s internal knowledge, RAG retrieves relevant information from a database or document corpus and feeds it into the model to produce more accurate, fact‑based responses. For a deeper dive, see our article on retrieval‑augmented generation.
Prompt engineering
The art of crafting questions and instructions for an AI model to get the best possible output. Good prompts provide context and clarify the desired format, tone and detail level. As the saying goes, garbage in, garbage out.
Context engineering
A broader term that includes prompt engineering plus the process of supplying background information—such as documents, images or data points—to help the AI generate more relevant answers. Think of it as briefing a human colleague before assigning a task.
Fine‑tuning
The process of training an existing AI model on a specific dataset to make it perform better on specialised tasks. Fine‑tuned models can understand industry jargon or company‑specific language but may be costly and risk overfitting.
Vector database
A type of database that stores data as numerical vectors (embeddings) rather than traditional rows and columns. Vector databases enable fast similarity search, which is critical for RAG systems that need to find relevant documents.
Agentic AI
AI systems designed to act autonomously to achieve goals. They use planning, reasoning and context to execute multi‑step tasks. Unlike rule‑based bots, agentic AI adapts to changing circumstances..
Local LLM
A language model deployed on your own infrastructure or within a private cloud. Local LLMs offer greater control over data, lower latency and more customisation options than public models. Read about why businesses are investing in private GPTs.
Hallucination
When an AI model generates incorrect or fabricated information with confidence. Techniques like RAG and human review help reduce hallucinations by grounding responses in real data.
AI power‑user
A term describing individuals who extensively use AI tools to enhance their productivity and creativity. Research shows that companies that invest in training to nurture AI power‑users see transformative benefits.