Beginner glossary
This glossary is for anyone getting into AI and tripping over the same terms again and again. Each definition is short enough to skim and clear enough to actually use.
Use it with the newsletter
Each ACC Network issue will call out one word or term of the day in a compact definition bubble, then point back here for the full beginner reference.
Term
An AI system trained to understand and generate text.
A large language model is an AI model trained on huge amounts of text so it can answer questions, summarize information, write drafts, and follow instructions in natural language.
Why it matters
This is the engine behind tools like ChatGPT, Claude, and Gemini, so beginners hear the term constantly.
Term
The instruction or question you give an AI tool.
A prompt is the input you type or speak to an AI system. It can be as short as one question or as detailed as a step-by-step request with context, constraints, and examples.
Why it matters
Better prompts usually lead to clearer, more useful answers, especially when you are learning.
Term
A small chunk of text an AI model reads or writes.
Models do not see words exactly the way humans do. They break text into tokens, which may be whole words, parts of words, punctuation, or spaces. Token counts affect cost and limits.
Why it matters
When a tool talks about price, context windows, or usage limits, tokens are usually the unit being counted.
Term
The amount of information a model can consider at once.
A context window is the maximum amount of text, instructions, and uploaded content a model can keep in mind in a single interaction.
Why it matters
A larger context window makes it easier to work with long documents, codebases, and multi-step tasks.
Term
The moment an AI model is actually generating an answer.
Inference is what happens after the model has already been trained and is now using what it learned to respond to a prompt, classify data, or generate an output.
Why it matters
Many AI products charge for inference, and speed during inference shapes how responsive a tool feels.
Term
Training an existing model further on a narrower task or dataset.
Fine-tuning takes a base model and teaches it additional behavior using targeted examples, often so it performs better on a specific style, domain, or workflow.
Why it matters
Beginners hear this term when deciding whether prompt engineering alone is enough or whether a custom model setup is worth it.
Term
A numerical representation of meaning used for search and matching.
An embedding turns text, images, or other content into numbers that capture semantic meaning, which makes it possible to compare similarity and retrieve related information.
Why it matters
Embeddings power AI search, recommendation systems, and retrieval workflows behind many modern assistants.
Term
An AI system that can take multiple steps toward a goal.
An agent is a workflow or application that uses a model to plan, decide, and often call tools or APIs across several steps instead of only returning one answer.
Why it matters
A lot of current AI products market themselves as agents, so knowing the difference between an agent and a chatbot saves confusion.
