Resources
AI Glossary
The handful of terms you actually hear — defined in plain language, so you can follow the conversation without pretending to know the jargon.
- Artificial Intelligence (AI)
- Software that learns a task by studying examples, rather than relying on a programmer to spell out exact instructions for every situation.
- Algorithm
- A set of step-by-step instructions a computer follows to accomplish something. Most software is built on algorithms; AI simply uses algorithms that learn from data.
- Machine Learning
- The branch of AI where the software improves at a task the more examples it sees — spotting patterns on its own instead of being told the rules.
- Large Language Model (LLM)
- A type of AI trained on enormous amounts of text so it can predict and generate language. ChatGPT and similar tools are powered by LLMs.
- Prompt
- The instruction or question you give an AI tool. How clearly you write the prompt largely determines how useful the answer is.
- Chatbot
- A program that converses with people in plain language. Modern chatbots are powered by AI; older ones followed rigid, pre-written scripts.
- Hallucination
- When an AI produces an answer that sounds confident and correct but is actually wrong. It is the main reason a human should always check important output.
- Training Data
- The collection of examples an AI studies to learn its task. The quality and breadth of this data shapes what the AI can and can’t do well.
- Generative AI
- AI that creates new content — text, images, audio — rather than only sorting or predicting from existing options.
- Automation
- Having software carry out a task on its own, without a person stepping in each time. Automation can use AI, but the two are not the same: a simple email auto-reply is automation, not AI.
- Fine-tuning
- Taking a general AI model and giving it additional, more specific training so it performs better on a particular kind of task or industry.
- Inference
- The moment the AI actually answers your question — putting what it learned during training to work in real time.
- Artificial General Intelligence (AGI)
- A hypothesized AI that can match a human across any intellectual task, not just the narrow one it was built for. As of today, AGI does not exist; it remains a research goal, not a product.
- Artificial Superintelligence (ASI)
- A further-off concept: an AI that surpasses human capability across every domain. It is the subject of serious debate and concern, but it is not something on the market or in your tools.
- AI Agents
- Tools that take a goal, break it into steps, and act on your behalf — reading a screen, sending an email, updating a record — rather than only answering a single question. They are newer and still imperfect — useful, but double-check their work.
- Copilot
- A branded term for an AI assistant embedded inside another program — sitting beside you in your email, spreadsheet, or code editor and helping as you work.
- Tokens
- The small chunks of text (a word, part of a word, or punctuation) that an AI model reads and generates. Usage limits and costs are often measured in tokens rather than words.
- Context Window
- How much text an AI can “hold in mind” at once — its memory for a single conversation or document. Hand it more than the window allows and it starts to forget the earlier parts.
- Retrieval-Augmented Generation (RAG)
- A technique where the AI is pointed at your own documents first, then answers from those — so its replies are grounded in your information instead of only its general training.
- Multimodal
- A model that can work with more than one kind of input or output — text, images, audio, or video — rather than text alone.
- Open-Source Models
- AI models whose underlying code is shared publicly so anyone can run or adapt them. They are an alternative to the closed, commercial tools most people first encounter.
- Model Context Protocol (MCP)
- A newer open standard that lets AI tools connect to your data sources and apps in a consistent way — the plumbing that helps an AI actually reach your files and tools rather than guess about them.
Hear a term that isn’t here? Ask me on a call and I’ll define it for your situation.