Reference

Glossary

Every word here shows up somewhere else in the toolkit — usually underlined, so you can tap it in context. Nobody expects you to know these already.

AI Foundations

AI (Artificial Intelligence): Technology that enables computers to perform tasks that typically require human intelligence, such as recognising patterns, understanding language, making predictions or generating content.
AI Tool: An application or service that uses artificial intelligence to help users perform a task, such as writing, researching, creating images, translating or analysing information.
Algorithm: A set of instructions or rules that a computer follows to solve a problem, process information or make a decision.
Data: Information that can be collected, stored, processed or analysed by people or computer systems. Data can include text, numbers, images, audio, video and other forms of information.
GenAI (Generative AI): A type of AI that can create new content, such as text, images, audio, video or code, in response to instructions and based on patterns it learned from huge amounts of data. ChatGPT and Claude are GenAI. It is designed to mimic human intelligence.
Large language model (LLM): The engine behind chat assistants. It predicts the next most likely piece of text, which is why it sounds confident even when it's wrong.
Training Data: The information used to train an AI model so that it can learn patterns and relationships and produce outputs.

Working with AI

Accuracy: How correct or factually reliable an AI output is.
AI Output: The response or content produced by an AI system, such as text, an image, a recommendation, code or audio.
Confidence: How certain an AI system appears to be about its response. An AI can sound very confident even when its answer is wrong.
Context: The information provided to an AI system that helps it understand what a request means and how it should respond.
Context window: How much text a model can hold in mind at once. Outside that window, it simply doesn't know — it has no memory of your business by default.
Hallucination: When an AI generates information that is incorrect, invented or unsupported but presents it as though it were true. It happens because the model predicts likely words, not verified truth.
Human in the Loop: An approach where a person remains involved in reviewing, approving, correcting or making decisions about an AI system's outputs or actions before it affects anyone. A human stays accountable for the final call.
Knowledge Cutoff: The point in time up to which information was included in the training or knowledge available to an AI system. Some AI systems can access more recent information through tools such as web search.
Prompt: The instruction, question or information a person gives an AI system to tell it what they want it to do. The clearer the role, context, task, format and constraints, the more useful the answer.
Prompt Engineering: The practice of designing and refining prompts to help an AI system produce more useful, accurate or relevant results.
Token: The chunks of text a model reads and writes — roughly a short word or part of a word. Limits on tokens are why very long documents get truncated.
Verification: The process of checking information produced by AI against reliable sources or other evidence before relying on it.

Data, Privacy and Personal Information

Digital Footprint: The record of information and activity that a person leaves behind when using digital services, websites, apps and online platforms.
Personal Data: Information that can identify a person, either directly or when combined with other information. This can include a name, email address, location, photograph or online identifier.
Privacy: The right to control information about yourself and to have your personal information handled appropriately.
Sensitive Data: Personal information that could create greater risks for someone if it were exposed or misused. Examples can include health, financial, biometric or other highly private information.

Bias, Ethics and Responsible AI

Accountability: The responsibility of people or organisations for the decisions, actions and consequences associated with an AI system.
AI Ethics: The principles and considerations used to think about whether AI is being developed and used in ways that are fair, safe, responsible and beneficial to people and society.
AI Governance: The policies, processes, rules and responsibilities used to manage how AI is developed, implemented and used responsibly.
Bias: A tendency to favour, disadvantage or represent certain people, groups or perspectives in a particular way. Bias can exist in data, algorithms, AI systems or how people use them.
EU AI Act: The European Union's law on AI. It sorts AI uses into risk tiers — from minimal risk through transparency obligations and high-risk duties to outright bans.
Transparency: Being open about how an AI system is being used, what it does, what information it relies on and what its limitations are.

AI, Manipulation and Information

AI Manipulation: The use of AI to deliberately alter, generate or present information in a way that influences how people think, feel or behave.
Deepfake: AI generated or AI manipulated audio, video or images that make it appear that a real person said or did something they did not actually say or do.
Disinformation: False or misleading information that is deliberately created or shared with the intention of deceiving or manipulating people.
Misinformation: False or inaccurate information that is shared without necessarily intending to deceive people.