AI is everywhere, from chatbots to image generators. It’s in our homes, our schools and being seen more in our everyday lives. But how much do we really understand about how AI works? We’ve gathered 22 key AI terms for you, but let’s be honest: most people won’t remember them all.
Five AI Terms to Know First
Before we dive into the full list, here are the five you really need to know. These are the basic terms that everyone should know when stepping into the world of AI.
- Artificial Intelligence (AI): The umbrella term for computer systems that perform tasks requiring human-like intelligence.
- Machine Learning (ML): How many AI systems learn patterns from data rather than following fixed instructions.
- Generative AI: Models that create new content such as text, images, audio, or code.
- Large Language Models (LLMs): Powerful text models behind tools like ChatGPT that understand and generate natural language.
- Hallucination: When an AI produces confident but incorrect information, reminding us to verify important outputs.
Together, these five give you the essentials: what AI is, how it works, why it’s exciting, and where to be cautious.
The Building Blocks of AI
AI starts with a few core ideas that explain what it is, how it works, and why it matters today.
- Artificial Intelligence (AI) – Computers performing tasks that usually require human intelligence.
- Machine Learning (ML) – A method of teaching systems to learn from data rather than direct instructions.
- Deep Learning – A type of ML that uses many layers of networks for complex tasks like vision and speech.
- Neural Network – A web of connected “neurons” that process data step by step to reach a decision.
- Generative AI – Models that create new content such as text, images, or music.
- AI Models – The general term for any trained system that can perform a task using data (could be language, images, or even detecting fraud).
How AI Models Work
Inside every AI Model are mechanisms that determine what it knows, how it processes information, and its limits.
- LLM (Large Language Models) – A specific kind of AI model trained on massive text datasets to understand and generate natural language (e.g., ChatGPT, Gemini, Claude).
- Parameters / Weights – The learned numbers that shape how a model makes decisions.
- Tokens – Small text chunks that models read and produce when generating language.
- Embeddings – Number-based representations that capture the meaning of words and concepts.
- Context Window – The maximum amount of text a model can “remember” at once.
- Knowledge Cut-off – The data after which the model has no training data.
Learning and Adapting
Different approaches allow AI systems to learn, improve, and adapt to tasks in varied ways. For example, AlphaGo is an AI program developed by DeepMind that mastered the complex board game Go by combining deep neural networks with reinforcement learning through millions of self-played games. AlphaGo famously beat Lee Sedol, one of the world’s top Go players, in March 2016, winning 4 to 1 in a five-game match.
- Reinforcement Learning (RL) – Learning by trial and error, guided by rewards and penalties. AlphaGo learned by playing against itself and other Go playing professionals.
- Reinforcement Learning with Human Feedback (RLHF) – Using human feedback to fine-tune AI behaviour for helpfulness and safety.
- Supervised Learning – Training with labelled examples to show the model what “correct” looks like.
- Unsupervised Learning – Finding hidden patterns in unlabelled data.
- Prompt Engineering – Crafting clear instructions to guide better AI outputs
Using AI Responsibly
AI is powerful, but it can also make mistakes; it raises questions about safety, trust, and responsible use.
- Hallucination – When AI generates confident but incorrect information.
- Bias – Hidden unfairness that comes from the data AI learns from.
- Temperature – A setting that controls whether AI outputs are predictable or creative.
- RAG (Retrieval-Augmented Generation) – Combining AI with trusted sources to improve accuracy.
- Edge / On-device AI – Running AI locally on devices for speed and privacy.
AI may seem complicated, but it can be a great help in our day to day lives, education and development. By knowing these 22 key terms, you can cut through the jargon and begin to understand the subject with confidence.
By Tara Kemp