The glossary.
The terms behind AI, tech and robotics news, in plain English. Each links to a full, sourced explainer.
A
- AGI(AGI (artificial general intelligence))
- AGI means AI that can match or beat humans across most thinking tasks, not just one. It's the stated goal of labs like OpenAI, but there's no agreed definition and experts disagree sharply on when, or whether, it will arrive. Read more →
- AI agent(AI agents)
- An AI agent is an AI system that decides its own next steps and uses tools to get a task done, instead of just answering one question. That makes agents powerful, and harder to control. Read more →
- AMR(autonomous mobile robot (AMR))
- An autonomous mobile robot (AMR) is a wheeled robot that moves goods or does tasks on its own, using sensors and a map of the building to find its way. Unlike older guided vehicles on fixed tracks, it plans its own route around obstacles. Read more →
C
- chipmaker(semiconductor chip)
- A semiconductor is a material, usually silicon, that can either conduct or block electricity. Chips pack billions of tiny switches made from it onto a small piece of silicon, and they power everything from phones to AI data centers. Read more →
- cobot
- A cobot, or collaborative robot, is a robot designed to work safely right next to people, without the safety cages that surround traditional industrial robots. Read more →
- context window
- A context window is how much text an AI model can see and work with at once: your instructions, the conversation so far, any files, and its own reply. It's measured in tokens, small chunks of text of roughly four characters each. Read more →
D
- deep learning(neural network)
- A neural network is an AI model made of many simple connected units, loosely inspired by neurons in the brain. Nobody writes its rules: it learns patterns from examples by gradually adjusting the strength of its connections. Read more →
- deepfake
- A deepfake is a video, photo or audio clip made or altered with AI so it looks or sounds real. It can make a real person appear to say or do something they never did, or invent a face or voice that belongs to no one. Read more →
- delivery drone(How do delivery drones work)
- Delivery drones are small electric aircraft that fly packages from a store or warehouse to your home, mostly on autopilot. Trained staff watch over whole fleets at once, and the package is usually lowered on a line into your yard. Read more →
E
- exosuit(exoskeleton)
- An exoskeleton is a wearable frame that supports or boosts the work of a person's joints and muscles. Powered versions help some people with paralysis stand and walk; unpowered ones take strain off workers who lift or reach overhead. Read more →
F
- fine-tune(fine-tuning an AI model)
- Fine-tuning takes an AI model that has already learned from huge amounts of general data and trains it a bit more on a small set of examples for one task or field. The result is a customized model, without building one from scratch. Read more →
G
- GenAI(generative AI)
- Generative AI is artificial intelligence that creates new content, such as text, images, audio, video or code, from a prompt. It learns patterns from huge amounts of existing data and uses them to produce something new. Read more →
- GPU(GPU, and why does AI need them)
- A GPU (graphics processing unit) is a chip that does huge numbers of simple calculations at the same time. Built to draw 3D graphics, that parallel power now makes GPUs the main engine for training and running AI. Read more →
H
- hallucinate(AI hallucinations)
- An AI hallucination is when a chatbot confidently states something false or made up while sounding completely believable. It's a side effect of how these models generate text, not a sign that the AI is lying. Read more →
- humanoid(humanoid robot)
- A humanoid robot is a machine built in the shape of a person, with a torso, head, arms and usually legs, so it can work in spaces and with tools designed for humans. Read more →
- hyperscaler(AI data center)
- An AI data center is a huge building full of specialized computers, mostly GPUs, that train AI models and answer requests to them. It works like a normal data center but needs far more electricity, faster connections and much stronger cooling. Read more →
I
- inference(AI inference)
- AI inference is when a trained model is put to work on something new, like answering your prompt or recognizing a photo. Training is how a model learns; inference is how it gets used, every single time someone asks it something. Read more →
L
- levels of self-driving(levels of self-driving mean)
- Self-driving is graded on six levels, from 0 (you do everything) to 5 (the car drives anywhere by itself). The big line is between Level 2, where you must supervise, and Level 3 and up, where the system is driving. Read more →
- lidar
- Lidar (light detection and ranging) is a sensor that fires laser pulses and times how long they take to bounce back. Millions of those measurements build a detailed 3D map of the surroundings, used by self-driving cars, robots, mapmakers and even phones. Read more →
- LLM(large language model (LLM))
- A large language model is an AI system trained on huge amounts of text to predict the next word. That simple skill, at massive scale, lets it write, summarize, translate and answer questions. Read more →
M
- machine learning
- Machine learning is a branch of AI where software learns patterns from example data instead of following rules a person writes by hand. Once trained, it uses those patterns to make predictions or decisions about new data. Read more →
- MCP(MCP (Model Context Protocol))
- MCP, the Model Context Protocol, is an open standard that lets AI apps like Claude, ChatGPT and Gemini connect to outside tools and data in one common way. Anthropic created it in 2024, and it's now run under the Linux Foundation. Read more →
- multimodal(multimodal AI)
- Multimodal AI is AI that can work with more than one type of data, such as text, images, audio and video. It can look at a photo and answer in words, or listen to speech and reply out loud. Read more →
O
- open-weight(open-weight AI models)
- An open-weight AI model is one whose trained parameters (the 'weights') are published for anyone to download and run. That's not the same as fully open-source AI, which also shares training data and code. Read more →
P
- physical AI
- Physical AI is artificial intelligence that runs machines in the real world, like robots and self-driving cars. It senses its surroundings, understands what's happening and then moves or acts, instead of only producing text or images. Read more →
R
- RAG(RAG (retrieval-augmented generation))
- RAG, or retrieval-augmented generation, is a way for an AI chatbot to look up relevant information in outside sources, like a company's documents, before it answers. That makes answers more current, more specific and easier to check. Read more →
- reasoning model(AI reasoning models)
- An AI reasoning model is a language model that works through a problem step by step before it answers: breaking it down, trying approaches and checking its work. That makes it slower and pricier, but more reliable on hard, multi-step problems. Read more →
- robotaxi
- A robotaxi is a ride-hailing car that drives itself with no human driver. Today they run only inside mapped service areas, in a handful of cities in the US and China. Read more →
T
- TPU
- A TPU (tensor processing unit) is a chip Google designed specifically to speed up machine learning. It's built for the huge matrix math that neural networks run on. Read more →