You do not need to understand how an engine works to drive a car, but it helps to know the difference between the accelerator and the brake. The same is true of AI. Here are the terms you will actually meet, and what they mean for a business.
Artificial intelligence (AI)
The broad idea: software that does things we used to think needed a person — recognising a face, understanding a sentence, spotting a pattern. Your email spam filter has been "AI" for twenty years. The recent excitement is about one branch in particular.
Generative AI
Software that produces things — text, images, code, summaries — rather than only sorting or recognising them. When people say "AI" today, they usually mean this.
Large language model (LLM)
The engine inside generative AI for text. It has read an enormous amount of writing and learned, statistically, what words tend to follow other words — well enough to draft an email, summarise a report or answer a question. It does not "know" things the way a person does, which is why it can be confidently wrong.
ChatGPT, Copilot, Gemini, Claude
Products built on top of models. ChatGPT is OpenAI's consumer product. Microsoft Copilot is Microsoft's assistant, built into Windows, Microsoft 365 and the Edge browser. The important distinction for a business is not which one writes better emails; it is where your data goes and what the tool can see — covered in our article on Copilot versus free tools.
Prompt
Whatever you type or say to the assistant. Better prompts get better results — "summarise this thread and list the decisions" beats "help".
Hallucination
The polite term for the model making something up: a statistic, a case, a policy that does not exist. It is not lying; it is guessing fluently. Everything factual gets checked by a person.
Agent
An AI set up to carry out a multi-step task — check a mailbox, look something up, draft a response, file it — rather than answer one question. Early days for most businesses, and covered in our advanced articles.
What this means for you
AI is a capable assistant that is fast, tireless and occasionally wrong. Used inside a well-managed Microsoft 365 environment with sensible rules, it saves real time. Used in a free tool with company data pasted in, it is a liability. The tool is the easy part; the environment and the rules are where we come in.
