A

  • Agentic AI describes workflows where models break goals into tasks, run tools, check intermediate outputs and adjust course with bounded autonomy.

  • An AI agent is software that uses a model to plan steps, call external tools, read results and pursue a set goal rather than replying once in chat.

  • AI alignment is the technical and design challenge of ensuring AI systems pursue human intentions safely without gaming metrics or causing side harm.

  • AI bias occurs when a model produces systematically skewed or unfair outcomes across groups due to unrepresentative training data or flawed labels.

  • An AI copilot is an assistant embedded inside everyday software that drafts, summarises or suggests next steps while the human user stays in control.

  • AI governance is the set of internal policies, audit trails, roles and approval checks an organisation uses to control how AI tools run in practice.

  • AI literacy is the practical ability to understand how AI tools work, frame tasks well, verify outputs critically and spot privacy or ethical limits.

  • AI safety studies how to test, constrain and monitor AI systems so failure modes, misuse risks and unexpected behaviours do not cause serious damage.

  • Answer engine optimisation (AEO) is structuring factual content and clear sources so AI search summaries and assistants cite your work accurately.

C

  • Computer vision is the field of AI enabling machines to inspect, classify and measure objects or defects inside photos, live video and medical scans.

  • A context window is the maximum amount of text, code or files an AI model can hold and process at one time across both your input and its response.

D

  • Deep learning is a form of machine learning using multi-layered neural networks to find complex patterns in raw text, audio, images and sensor inputs.

  • A deepfake is synthetic audio, video or imagery generated by AI to mimic a real person's voice, face or actions closely enough to mislead audiences.

E

  • The EU AI Act is European legislation regulating artificial intelligence by risk tier, banning certain practices and setting duties on high-risk uses.

F

  • Fine-tuning continues training an existing AI model on a smaller, curated dataset so it follows specific formats, tone or domain conventions reliably.

G

  • Generative AI refers to models that produce new text, code, images, audio or video from prompts by learning statistical patterns in training data.

H

  • An AI hallucination happens when a model generates fluent, confident output that is factually wrong, unsupported by sources or completely invented.

  • Human in the loop is a workflow design where a qualified person reviews, edits or approves AI outputs at high-stakes checkpoints before action runs.

L

  • A large language model (LLM) is an AI system trained on text to predict tokens in sequence, letting it draft, summarise, translate and write code.

M

  • Machine learning is a branch of computing where algorithms learn patterns from historical data to make predictions or decisions without manual rules.

  • Model Context Protocol (MCP) is an open standard connecting AI assistants safely to external databases, files and workplace tools without custom glue.

  • Multimodal AI refers to models that accept and connect multiple data types at once, including text, images, audio, video and structured spreadsheets.

N

  • Natural language processing (NLP) is the branch of AI enabling computers to parse, search, classify, translate and generate written or spoken language.

  • A neural network is a mathematical model made of connected layers of weighted nodes that adjust during training to map raw inputs to useful outputs.

O

  • An open-weight model is an AI model whose trained numerical parameters are publicly downloadable, letting teams inspect, fine-tune and self-host it.

P

  • Prompt engineering is structuring instructions, context, examples and constraints clearly so an AI model produces reliable, verifiable work output.

R

  • Responsible AI is the operational practice of building and deploying AI with clear accountability, fairness checks, data privacy and human oversight.

  • Retrieval-augmented generation (RAG) fetches relevant passages from trusted documents before a model answers, grounding replies in real reference text.

S

  • Synthetic data is artificially generated information created by software or simulations to train and test AI systems when real records are sensitive.

T

  • A token is the basic chunk of text, code or data an AI model processes, often equal to a word, part of a word, a number or a punctuation character.

  • A transformer is a neural network architecture using attention mechanisms to weigh relationships across sequences, powering modern language AI models.

V

  • Vibe coding is building software mainly by prompting AI coding tools in plain English and iterating on behaviour rather than writing every line manually.

See these concepts inside real industries

One issue every two weeks examining how AI actually changes work across fifteen sectors. Free, written by students.