algorithmic progress
Improvement in AI capability per unit of compute due to better algorithms, measured as reduction in compute needed to reach a fixed benchmark.
As defined in Carl Shulman (Pt 1) — Intelligence explosion, primate evolution, robot doublings, & alignment
Other definitions
Improvements in AI techniques and algorithms (beyond just adding more compute). Measured in OOMs—currently ~0.5 OOM/year. Represents how much more capability per unit of compute is achieved through better methods.
Leopold Aschenbrenner — 2027 AGI, China/US super-intelligence race, & the return of history
Where it comes up · 6
- Dwarkesh Patel Podcast - Paul Christiano – We could be 12 months away from AIs that can overthrow the government5 claims
- What Will AI Look Like in 2027? | Interview2 claims
- Leopold Aschenbrenner — 2027 AGI, China/US super-intelligence race, & the return of history2 claims
- AI Trends 2025: Mary Meeker Deck Deep Dive Part 11 claim
- Carl Shulman (Pt 1) — Intelligence explosion, primate evolution, robot doublings, & alignment1 claim
- Dwarkesh Patel Podcast - The avalanche that hit math is coming for AI development1 claim
Most important claims · 10 of 12Search all claims for “algorithmic progress” →
- People tend to underestimate the rate of change and overestimate the difficulty of being in the trenches of a field. Being in close focus on immediate technical challenges (debugging current models, fixing data issues) reduces visibility into macro trends (exponential capability scaling, approaching AGI) and makes algorithmic progress seem less impressive than it actually is.Leopold Aschenbrenner — 2027 AGI, China/US super-intelligence race, & the return of history
- As AI development itself becomes increasingly automated, automation increases the quality and quantity of labor applied to AI R&D, so capability changes that would normally take four to six years could be compressed into a very short period.Dwarkesh Patel Podcast - Paul Christiano – We could be 12 months away from AIs that can overthrow the government
- AI developers should pause further algorithmic progress on pre-training and RL efficiency until their security posture can defend their innovations and they can tell whether they control the systems they built.Dwarkesh Patel Podcast - Paul Christiano – We could be 12 months away from AIs that can overthrow the government
- Automating the full AI research process creates a 25x acceleration in algorithmic progress, because the research loop can be executed continuously without the bottleneck of human researchers, while compute scaling remains constant.What Will AI Look Like in 2027? | Interview
- Once AGI/drop-in remote workers arrive and can be scaled to productivity levels, the next logical automation target is AI research and engineering itself. Once AI researchers are automated, you can run millions or billions of human-equivalent automated AI researchers on GPU fleets, enabling a decade's worth of ML research progress to be compressed into a single year, triggering a 10x speedup in research velocity and potentially a jump to vastly superhuman AI within 1-2 years.Leopold Aschenbrenner — 2027 AGI, China/US super-intelligence race, & the return of history
- 200% annual growth over nine years of compute gains from better algorithms, combined with raw compute scaling, demonstrates that future AI improvements will likely come increasingly from algorithmic progress and better use of existing compute rather than just scaling raw compute or pre-training data over the next 5-10 years.AI Trends 2025: Mary Meeker Deck Deep Dive Part 1
- Data from METR and Anthropic is consistent with the pace of AI progress having roughly doubled over the last 6-12 months, even after accounting for diminishing returns and compute bottlenecks, though the underlying evidence is not public.Dwarkesh Patel Podcast - Paul Christiano – We could be 12 months away from AIs that can overthrow the government
- Once superhuman coders exist, they would be deployed to automate the full AI research process within roughly six months (halfway through 2027 in the scenario), achieving 25x faster algorithmic progress while compute scaling remains flatWhat Will AI Look Like in 2027? | Interview
- For pretraining over the last six years, more than three times as much algorithmic progress has come from data improvements as from model improvements, according to research Patel did with Jerry Han.Dwarkesh Patel Podcast - The avalanche that hit math is coming for AI development
- Machine learning research probably lags mathematics by roughly one year in AI capability, so if math is advancing this quickly, Christiano expects roughly ten years of machine learning ideas could arrive in the near future.Dwarkesh Patel Podcast - Paul Christiano – We could be 12 months away from AIs that can overthrow the government