AI Glossary
Essential AI terms and definitions to help you navigate the world of artificial intelligence and understand the concepts shaping the future of work.
Each term has a plain-English definition and an example of where you would actually run into it at work.
Agent
An AI that can take action for you. Instead of just answering questions, agents connect to tools, apps, or data sources to get things done, like scheduling a meeting, pulling info from your email, or writing a draft and sending it.
Example: You set up an agent that watches for new inquiry emails, pulls the sender's company details, and drafts a reply in your voice for you to approve. You still decide what gets sent.
Context
The information an AI uses to understand your request. This might include the conversation so far, documents you've connected, or even data from your tools. The better the context, the better the AI's response.
Example: An approved AI tool can draft a proposal more usefully when you give it the client's brief, an example you're allowed to share, and a clear description of your services. Review the draft before sending it.
Fine-tuning
Training an existing AI model on specialized data so it learns a particular style, skill, or domain. Example: fine-tuning a model on legal documents so it becomes better at drafting contracts.
Example: Most professionals never need this. If a vendor says their tool is fine-tuned for your industry, ask what it does better than a general model given good instructions and examples.
Generative AI
The type of AI that can create new content, text, images, code, music, video, rather than just analyze existing data. Examples include ChatGPT (text), Midjourney (images), and Sora (video).
Example: When you ask Claude or ChatGPT to turn meeting notes into a follow-up email, or use Nano Banana to make a slide image, you are using generative AI.
Hallucination
When an AI makes up information that isn't true but presents it confidently. It's like the AI's version of guessing. This is why human judgment and fact-checking are still essential.
Example: You ask for three statistics to support a slide and get a study that sounds real but does not exist. Before anything goes in front of a client, check every number and source yourself.
Memory
When an AI can remember things across sessions, like your preferences, projects, or past conversations, and use that information to improve future interactions. (Different platforms handle memory differently; some store everything, some let you control what's remembered.)
Example: After you tell ChatGPT once that you write in short paragraphs and work with coaching firms, later drafts start that way without being asked. Check the settings to see and edit what it has saved.
Prompting
The way you give instructions to an AI. Prompts can be simple ("write an email") or detailed ("draft a 3-paragraph email to my team about next week's project update, with a clear list of action items"). Good prompting = better output.
Example: Instead of "summarize this call," try: "Summarize this client call in five bullets, list every commitment I made with a date, and flag anything I should follow up on this week."
Vibe coding
Building apps, tools, or websites by chatting with AI instead of writing traditional code. You describe what you want in plain English, and the AI generates the working product. It's fast, creative, and accessible, even if you're not technical.
Example: You describe a simple intake form that scores new leads and emails you the good ones, and a tool like Lovable builds a working version you can test the same afternoon.