Also called: AGI · Last checked: 6 November 2026
In short: Artificial general intelligence (AGI) is a hoped-for kind of AI that could learn and do most tasks a person can, across many fields, without being built for each one. No system has reached it yet.
Table of Contents
How it works
- Today’s base. Companies start from large models that already handle language, images and code.
- Scaling up. They add more data, computing power and new training methods to widen what the models can do.
- Judging it. People test whether the system can match humans across a wide range of real work. There is no agreed test yet.
Where you’ll meet it
You meet the term in news about OpenAI and Google DeepMind, the makers of ChatGPT and Gemini, which both say AGI is their aim. You will not find it in any product you can buy.
What it means for your business
- Plan for today’s tools. Base decisions on what current AI can do now, tested on your own tasks.
- Treat AGI claims with care. Ask what a supplier’s product does today, whatever its long-term goals.
- Watch contracts and rules. AGI clauses already shape big supplier deals and may affect pricing and access.
Watch out for
- No shared definition. Companies define AGI in ways that suit them.
- Hype in sales pitches. AGI talk can distract from weak results in the product you are buying.
Example
OpenAI. OpenAI’s 2018 charter defines AGI as highly autonomous systems that outperform humans at most economically valuable work. In October 2025 OpenAI and Microsoft agreed new terms for their partnership. Under them, any AGI declaration by OpenAI must be verified by an independent expert panel. Microsoft said the revenue share it receives from OpenAI continues until the panel verifies AGI.
Artificial general intelligence vs superintelligence
| Artificial general intelligence | Superintelligence | |
|---|---|---|
| Ability | Matches people at most tasks | Far exceeds the best people at almost everything |
| Exists today | No | No |
| How they relate | The first goal | A possible later stage |
Frequently asked questions
When will AGI arrive?
Nobody knows. Some technology leaders predict within a few years, while many researchers expect decades or doubt it will happen. Forecasts vary because there is no agreed definition.
Would AGI replace managers?
It is too early to say, because AGI does not exist. Today’s AI handles parts of tasks, and people still set goals, judge trade-offs and carry responsibility.
How would anyone know AGI had been reached?
There is no agreed test. Some companies use expert panels or benchmarks, which are standard sets of tasks, but each group sets its own bar.