AGI
Artificial general intelligence: the hypothetical point where a system is broadly as capable as a person across most tasks, rather than good at one thing.
AGI stands for artificial general intelligence. The idea is a system that is not just excellent at coding, or chess, or summarising documents, but broadly capable across almost anything a person can do with a computer, including tasks nobody trained it for. Today’s models are impressively wide but still uneven: brilliant on one problem, confidently wrong on a neighbouring one.
There is no agreed test for when AGI has arrived, and that is not an oversight. Definitions vary from “can do most economically valuable work” to “matches an expert human on any cognitive task”, and the choice of definition changes the answer by years or decades. So when a lab executive says AGI is close, it is fair to ask what they mean by it before deciding whether to be excited or worried.
-
We Are Building an Invention Meant to Outgrow Its Inventors
-
Google reshuffles its AI leadership, and four of its most famous engineers walk out the door
-
Karpathy turned one paragraph of Tolkien into a 3D scene for ten dollars, and called it a vibe check
-
AMD trained a fully open model on its own chips and published everything except a commercial licence
-
OpenAI and Anthropic are arguing about a benchmark, and the argument is more useful than the scores
-
Amazon quietly stops developing most of its own AI models
-
An AI Helped Disprove an 87-Year-Old Math Problem Over the Weekend
-
DeepMind's Hassabis Wants an AI Referee, Modeled on Wall Street's Watchdog