Google DeepMind's Genewein maps four pathways from AGI to ASI
TL;DR
- Google DeepMind research scientist Tim Genewein argues superintelligence is a near-term trajectory, built on four compounding pathways past AGI.
- His essay projects that a cognitive workload costing $1 billion in compute today could cost just $1,000 in ten years.
- The piece draws on 'From AGI to ASI,' a technical report co-authored with 11 colleagues including DeepMind co-founder Shane Legg.
Tim Genewein, a research scientist at Google DeepMind, argues that superintelligence is not a speculative leap but a near-term trajectory built on pathways already in motion. His essay What comes after AGI?, published November 19 on the DeepMind Institute site, lays out four routes that could carry systems past human-level AGI: continued scaling of models and data, a shift beyond today's transformer-based paradigm, coordination of large populations of AGI instances into 'group agents,' and recursive self-improvement.
"Sheer scale can have a quality of its own," Genewein writes, arguing that even without new breakthroughs there is a major difference between slow, expensive human-level AIs and a world of billions of agents running much faster than humans. The essay projects that "a cognitive workload that can be performed by today's AI systems for $1 billion in compute tokens would cost just $1000 in ten years," and that the effective agent population "could scale up by more than a thousand-fold by 2030, and more than a billion-fold within ten years."
The essay defines ASI as a system that "consistently outperforms large groups of human experts on tasks that could take them years to complete." It draws on a technical report, From AGI to ASI, co-authored with 11 colleagues including Shane Legg, Marcus Hutter, Joel Z. Leibo, Iason Gabriel and Allan Dafoe.
Both the essay and the hosting program carry an explicit distancing from Google: "DMI pieces are intended to serve as conversation starters, reflecting the author's ideas and research, and should not be read as Google's official view." The piece lands in the middle of a visible run of safety-focused coverage; our safety tracker has logged 375 stories in the last 90 days.
Originally reported by institute.deepmind.com
Read the original article →Original headline: Google DeepMind's Tim Genewein Essay Argues AI Trajectory Now Points From AGI to ASI Within a Decade