Click here to download a printable PDF
Agentic AI: AI that operates with a sense of autonomy — it can take action, make choices, and pursue goals, but still within limits set by its design.
AI (Artificial Intelligence): Computers doing things that usually need human brainpower – like understanding language, recognising faces, or playing chess.
AI agent: An autonomous AI system that can plan, make decisions, and act to achieve goals, often based on real-time input.
AI tools: Software that uses artificial intelligence to perform tasks like writing (e.g. ChatGPT) or image creation (e.g. DALL·E).
Algorithm: A step-by-step set of rules a computer follows to solve a problem – a bit like a recipe, but for machines.
API (Application Programming Interface): A set of rules that lets different software systems communicate with each other.
Chatbot: A program that talks (or chats) with you – like ChatGPT. It can answer questions, help you write things, or even have a bit of banter.
Context window: The amount of text an AI model can “remember” at once. A larger context window helps it follow longer conversations or documents.
Data: The raw information AI learns from – like photos, emails, or numbers in a spreadsheet.
Deep Learning: A clever type of machine learning using layers of ‘artificial neurons’.
Generative AI: AI that can make new stuff – like stories, images, poems, or music.
GPT (Generative Pre-trained Transformer): A type of AI model trained to understand and write human language.
Hallucination: When an AI generates confident-sounding answers that are factually wrong.
Machine Learning: Teaching computers to learn from data instead of fixed instructions.
Memory: ChatGPT’s ability to retain information between chats (only in the Plus version).
Model: The trained brain of the AI used to make decisions or predictions.
Prompt: The message or question you give to an AI to start the interaction.
Recommendation engine: An AI system that suggests things (like books or videos) based on your past behaviour or preferences.
Token: A small chunk of text (like a word or part of a word) used by AI to process language. Example: "ChatGPT is smart" = 5 tokens.
Tokenised: Text that has been split into tokens for an AI to understand and analyse.
Training Data: The examples an AI was shown to learn how to do its job.
Wicked problem: A complex issue with no clear solution, where fixing one part may worsen another. Often seen in ethics, climate, and AI.
Last updated: 5 June 2025