layman/terms
AI words

AGI

ay · jee · eye — Artificial General Intelligence

In shortAGI is the idea of an AI that could learn and do almost any thinking job a person can, instead of being good at just one kind of task.

Today's AI is like a brilliant assistant who only knows one trade. AGI would be the one who can pick up any trade you hand it.

Think about the assistants you’ve met. One is amazing with spreadsheets but couldn’t write a speech. Another writes beautifully but can’t fix a bike. Each is brilliant at their own trade.

Now imagine an assistant who could learn any job you handed them: accounts on Monday, a business plan on Tuesday, a science problem on Wednesday. That’s AGI, Artificial General Intelligence: an AI that could learn and do almost any thinking job a person can.

Where we are today

Why people disagree about it

There’s no single test for AGI. Some people say it means matching a person at almost every thinking task. Others say it means doing most paid work as well as people do. That’s why one expert can say it’s years away while another says it’s nearly here.

What it isn’t

AGI isn’t a robot from the films with feelings and a plan to take over. It’s about how many different kinds of thinking jobs an AI can learn. It doesn’t need a body and it doesn’t need to feel anything.

Where the comparison breaks down

A human assistant who learns every job still needs sleep, holidays and time to learn. An AGI could, in theory, be copied a thousand times and work on a thousand jobs at once. That’s a big part of why people take it so seriously.

How the words fit together

AI is the big umbrella. Today’s best general tools are built on LLMs. An AI agent is an AI that goes off and does tasks for you. AGI is the idea of an AI that could do almost any of those tasks as well as a person.

Watch out

AGI doesn't exist yet and nobody agrees on exactly when it would count as here. Treat any headline that says “AGI is here” or “AGI is next year” as an opinion, not a fact.

When you'll hear it

“We're getting closer to AGI.” AI keeps getting better at more kinds of tasks, so some people think one that can do almost anything isn't far off.

“That's narrow AI.” It's very good at one kind of job, like spotting faces or translating, but can't do much else.

“What happens after AGI?” People are asking how the world would change if machines could do most thinking work. Nobody knows for sure.

Say it in a meeting

“If AI keeps getting better at more kinds of work, which parts of our jobs should we plan for it to take on first?”