Large language model
A program that learned to use language by reading enormous amounts of text.
Full definition →The words behind the headlines, explained the way a patient friend would, with no jargon. 42 concepts across the basics, how models work, how people build with them, safety, and the business of AI.
A program that learned to use language by reading enormous amounts of text.
Full definition →The small chunk of text an AI reads and writes, often part of a word.
Full definition →How much text an AI can keep in mind at one time.
Full definition →The internal dials a model adjusts while learning, numbering in the billions.
Full definition →Training is teaching the model once; inference is using it every day.
Full definition →The instruction or question you give an AI to get a response.
Full definition →Hidden setup instructions that shape how an AI behaves in a chat.
Full definition →When an AI states something false with total confidence.
Full definition →An AI that can handle more than just text, like images, audio, or video.
Full definition →A setting that controls how predictable or creative an AI's answers are.
Full definition →The date after which an AI simply has not read anything.
Full definition →An AI that pauses to think through a problem step by step before answering.
Full definition →A big model split into specialists, only a few of which handle each question.
Full definition →Whether a model's inner settings are shared publicly or kept private.
Full definition →Training a small, cheap model to imitate a big, expensive one.
Full definition →Rounding a model's numbers to make it smaller and faster.
Full definition →Giving an already-trained model extra practice on a specific task.
Full definition →Turning words or images into numbers that capture their meaning.
Full definition →A standard test used to compare how good different AI models are.
Full definition →The most advanced, cutting-edge AI models available at any moment.
Full definition →Letting an AI look up real documents before it answers.
Full definition →An AI that can take actions and complete multi-step tasks on its own.
Full definition →An AI reaching outside itself to use calculators, search, or other software.
Full definition →A common standard that lets AI plug into apps and data sources.
Full definition →How an AI asks software to run a specific action with the right details.
Full definition →A special store that finds information by meaning rather than exact words.
Full definition →Safety checks that keep an AI's behavior within acceptable limits.
Full definition →Tests that measure whether an AI system is actually doing its job well.
Full definition →Sneaky hidden instructions that trick an AI into misbehaving.
Full definition →Making sure an AI's goals and behavior match what people actually want.
Full definition →Deliberately attacking an AI to find its weaknesses before others do.
Full definition →A trick that gets an AI to break its own safety rules.
Full definition →A hidden signal marking content as AI-generated.
Full definition →A short fact sheet describing what an AI model can and cannot do.
Full definition →Overseeing the powerful computer chips used to build advanced AI.
Full definition →A government body that tests and studies the risks of advanced AI.
Full definition →A doorway that lets other software talk to an AI model.
Full definition →You usually pay for AI by the token, for both what goes in and comes out.
Full definition →A cap on how many requests you can send an AI in a given time.
Full definition →AI that runs right on your phone or laptop instead of a distant server.
Full definition →A country building and controlling its own AI within its borders.
Full definition →Training is a giant one-time bill; inference is the ongoing running cost.
Full definition →