A position report from inside DeepMind, with Hutter (AIXI), Legg (Legg-Hutter score), Orseau, Leibo (multi-agent), Dafoe (governance), Graepel and others. It is not a technical paper proposing a method. It is a landscape map of what could come after human-level AGI, written by people who built much of the conceptual vocabulary the field uses.
"AGI" is not the finish line. It's a waypoint on a continuum whose theoretical endpoint (Universal AI / AIXI — see 05 - Universal AI (AIXI), informally) we already understand formally. What lies between AGI and that endpoint is the actual interesting territory, and we should map it now.
Most public discussion of AGI is shaped like "AGI day arrives → world changes." The report attacks this picture directly:
"...the image of a single transformative step change, caused by the introduction of human-level AGI into our society, could be inaccurate. More apt might be the prospect of a series of transformative societal changes caused by AI-enabled progress and breakthroughs across many areas of science and technology."
This is the bottom-line claim. Everything else in the paper is scaffolding to support it.
This is a Google DeepMind document published as research, not as a blog post. The conservatism shows: every claim gets a "but this may not happen" hedge. That hedging is itself a signal. They are not predicting ASI; they are saying the possibility cannot be ruled out and is worth preparing for at the level of "global, interdisciplinary endeavor."
The closing line from Turing they choose:
"We can only see a short distance ahead, but we can see plenty there that needs to be done."