Artificial General Intelligence (AGI) is a research goal aiming at achieving human-level AI technology within 5 to 10 years. An internal document from OpenAI shows a possible roadmap to achieve this. Let’s see how…

Artificial General Intelligence (AGI) refers to a type of AI that possesses the ability to understand, learn and perform any intellectual task that a human can. Unsurprisingly, the relentless quest for this general artificial intelligence is capturing the energies of researchers and the imagination of the public.

But how do we get there?

An internal OpenAI document containing a “roadmap” for achieving AGI leaked in July. This roadmap was then confirmed by Sam Altman (CEO of OpenAI) in September, so this is validated information. In this article, I will present the contents of this roadmap. It describes five milestones on the road to AGI.

It’s important to present this roadmap because there’s no doubt that the major players in AI foresee the existence of general-purpose AI in the relatively near future (5 to 10 years). I’ve already covered Anthropic CEO Dario Amodei’s article here. Sam Altman’s article The Intelligence Age, accessible here, is along the same lines, and Demis Hassabis, CEO of Google Deepmind, is also very ambitious, as he mentioned in a recent interview accessible here.