Will parallelization limits delay an intelligence explosion? | Epoch AI
Even with automated AI R&D producing millions of virtual researchers, progress may be capped by our ability to divide, coordinate, and recombine their work. Epoch AI on why parallelization technology is a missing parameter in intelligence explosion models.
Suppose we double all the labor, capital, land, and other rival inputs used to produce rival goods. At minimum, we can produce twice as much: we can always do the same things twice. But if we double all the inputs to research, there’s no analogous guarantee that technological progress proceeds twice as quickly. The two research teams could do the same work twice, but since ideas are nonrival, the repeated work will be mostly worthless. Having two copies of the same paper or blueprint, describing the same new algorithm or floorplan twice, is no more valuable than having one. So, more…
saved by
related reading
- Toward an O*NET for AI R&Depochai.substack.com
- How Should the US Prepare for Increasingly Automated AI R&D? | IFPifp.org
- Preparing for the Intelligence Explosionforethought.org
- Will AI R&D Automation Cause a Software Intelligence Explosion?forethought.org
- II. From AGI to Superintelligence: the Intelligence Explosion - SITUATIONAL AWARENESSsituational-awareness.ai
- AGI is still 30 years away — Ege Erdil & Tamay Besirogludwarkesh.com
- When AI builds itself \ Anthropicanthropic.com
- How quick and big would a software intelligence explosion be?forethought.org
- An Apple-Picking Model of AI R&D | Tom Cunningham – Tom Cunninghamtecunningham.github.io
- The software intelligence explosion debate needs experiments | Epoch AIepoch.ai
- Full automation of AI R&D probably yields a large speed up even without a software-only singularityblog.redwoodresearch.org
- Three Types of Intelligence Explosionforethought.org