NederlandsAGI
AGI (artificial general intelligence) is AI that can perform many different thinking tasks at a human level, rather than one fixed task. There is no agreed point at which something counts as AGI.
AGI stands for artificial general intelligence. It is AI that, like a human, can handle very different thinking tasks. A chess program can only play chess. A person can play chess, write a letter, sort out a bill and learn a new trade. AGI would have that breadth too.
How does it work?
AGI is not a product you can buy, but a goal that researchers are working towards. That is also why there is no agreed point at which a system may be called AGI. Some look at the breadth of what a model can do. Others look at whether it can do the work that currently needs people. That difference in definition explains why experts have such widely differing expectations.
The idea is old. In 1950, Alan Turing proposed replacing the question of whether machines can think with a practical test: can someone talking through a teleprinter tell a machine from a human? That proposal appeared in the journal Mind (link).
Why it matters
In the coming years the main question is how broad AI models become. A model that can tackle scientific questions on its own helps researchers in almost every field at once. That is one of the reasons AGI is so often in the news.
In the news you see progress in small pieces. On a new test, an AI model designed better AI experiments than the best of 24 human experts, an important part of the work through which AI can improve itself (link). Meanwhile, mathematicians debate whether AI really comes up with new ideas or mainly combines cleverly. An image with nails and an elastic band sharpens the question, and according to an economist, the next big leap for all of science lies precisely in that combining (link).
Such examples are not yet proof that AGI has arrived. They do show which questions are being investigated now: how broad is AI, and how good is it at coming up with something new?
Articles about AGI
-
Newest AI model can read the inside of other AI and write it itself
Researchers at MIT had OpenAI's GPT-6 Astra read and adjust the numbers that small AI networks are made of, without any training. It even wrote networks that play a computer game well. The researchers call these very early results.
-
Google DeepMind's Genie 4 builds interactive worlds, now with sound
Google DeepMind has presented Genie 4, an AI that creates a world you can control while you watch. What is new is that this world also has sound and conversations. A small group of testers is trying it out now; more users will follow.
-
What mathematicians think of OpenAI's 372 maths results
In early October, OpenAI put 372 results on open maths questions online. Mathematicians are impressed, but checking them will take months. What has been claimed, and what do experts think?
-
AI model designs experiments better than 24 experienced AI researchers
On a new test by P-Zero Research, Claude Opus 5.5 designs better AI experiments than the best of 24 human experts. The model needs less than half the computing time to do so and costs about one thirtieth as much.
-
Is AI just a remix machine? Mathematicians debate the rubber band around all knowledge
Now that AI finds mathematical proofs on its own, scientists are debating: is that truly new work, or clever combining of what was already known? An image of nails and a rubber band sharpens the question. According to an economist, that very combining holds the next big leap for all of science.
-
'Can machines think?' Alan Turing proposes an imitation game
In the journal Mind, Alan Turing proposes replacing the question of whether machines can think with a test: can an interrogator using a teleprinter tell a machine from a human?
