If you’ve been reading AI stories lately, you’ve probably run into words like AGI, LLM, hallucination, agent, and inference and quietly hoped someone would just explain them in normal English. You’re not alone. Too many people are now expected to understand AI jargon before anyone has actually explained it properly.
So here’s the simple version.
This is not a technical textbook. It’s a plain-English cheat sheet for some of the most common AI words people keep seeing online.
The AI Terms People Keep Hearing
AI
Short for artificial intelligence. This is the broad umbrella term for computer systems that can do tasks we usually think of as requiring human intelligence, like writing, recognising images, answering questions or making predictions.
AGI
Short for artificial general intelligence. This is the idea of AI that could think and perform across a wide range of tasks at a human-like or beyond-human level. The tricky part is that people still don’t agree on exactly what counts as AGI, which is why the term often creates more heat than clarity.
LLM
Short for large language model. This is the kind of AI model behind tools like ChatGPT, Claude and Gemini. A large language model learns patterns from huge amounts of text and then predicts what words should come next when you give it a prompt.
Hallucination
This is the AI industry’s word for when a model makes something up and presents it like it’s true. That could mean a fake fact, a made-up quote, a fake source, or a confident answer that sounds right but isn’t. TechCrunch’s glossary describes hallucination as AI generating information that is incorrect.
AI agent
An AI agent is usually described as a system that can do more than just answer one question. It can take actions, carry out steps, use tools, and work toward a goal on your behalf. In simple terms, a chatbot talks. An agent is supposed to do.
Inference
This is the process of actually running an AI model to get an output. Training is when the model learns. Inference is when it uses what it learned to answer your prompt, generate an image, summarise a document, or do some other task. TechCrunch’s glossary includes inference as one of the core terms readers keep seeing.
Neural network
A neural network is one of the main building blocks behind modern AI. It’s a model architecture loosely inspired by how the brain processes information, although it is far simpler than an actual brain. You don’t need to understand the maths to know the basic point: this is part of the machinery behind many modern AI systems.
Prompt
A prompt is just the instruction you give the AI. It can be a question, a request, a command, or a block of text telling the system what you want. If you type “Write me a short email” or “Summarise this PDF,” that’s a prompt.
Multimodal
This means an AI system can work with more than one kind of input or output, such as text, images, audio or video. A multimodal model might read an image, answer questions about it, and then generate text in response.
Diffusion
Diffusion is a term often used in image generation. It refers to the type of model architecture behind many AI image tools. The basic idea is that the model learns how to turn visual noise into a finished image step by step. TechCrunch includes diffusion as one of the terms people are increasingly seeing in AI discussions.
Distillation
Distillation is when a smaller AI model learns from the outputs of a larger one. The goal is to make a model that is cheaper and faster while still keeping some of the larger model’s useful abilities. TechCrunch’s glossary highlights distillation as a common AI term worth understanding.
Chain of thought
This usually refers to an AI breaking a problem into intermediate reasoning steps instead of jumping straight to an answer. The concept matters because it can improve performance on some tasks, even if users do not always see all the internal reasoning.
Fine-tuning
Fine-tuning means taking an existing model and training it further for a narrower purpose, such as legal writing, coding, customer support, or medical note summarising. It’s one way companies try to make models more specialised.
RAG
Short for retrieval-augmented generation. This means an AI system pulls in outside material, such as documents or databases, before generating an answer. In plain English, it is one way of helping the model rely on real source material instead of just guessing from memory.
API
Short for application programming interface. In the AI world, this usually means a way for developers to connect an app or website to an AI model so the model can do work behind the scenes.
Coding agent
This is a more specific version of an AI agent. It is designed to help write, test, debug and sometimes run code more autonomously. TechCrunch’s glossary includes coding agents as one of the newer ideas moving into mainstream AI coverage.
Why These Terms Matter
A lot of AI jargon sounds more intimidating than it really is. Once you strip away the buzzwords, many of these ideas are surprisingly simple:
- a prompt is just an instruction
- a hallucination is a made-up answer
- an LLM is the kind of model behind chatbots
- an agent is AI that takes actions, not just chats
- AGI is the still-debated dream of more general machine intelligence
That’s why glossary-style guides are useful. They stop AI coverage from feeling like a private club where everyone pretends to know what every acronym means. TechCrunch explicitly framed its glossary as a guide to the words and phrases readers keep running into as AI terminology spreads.
The Bigger Point
The fact that people now need an AI glossary at all tells you something important: AI has become mainstream enough that the language around it is spilling into everyday life. When terms like hallucination, agent and inference start showing up in normal news stories, product launches and workplace conversations, plain-English explanations stop being optional. They become part of basic digital literacy.
Why this matters for Australia
Australian readers are dealing with the same jargon overload as everyone else. AI terms are turning up in school discussions, workplace tools, news stories, government debates and everyday apps. If people are going to keep hearing words like AGI, hallucination and AI agent, they need simple explanations that don’t assume a computer science background.
That’s why this kind of glossary matters. It helps readers tell the difference between a useful concept, a vague buzzword and a piece of marketing hype.
The bigger takeaway is simple: as AI gets more mainstream, understanding the language around it is becoming a basic skill, not just a niche tech hobby.
Source: TechCrunch
