Astral Codex Ten

Davidson

5 min read

Davidson

Article

Davidson is a recurring person in the Astral Codex Ten archive, appearing 2 times across 2 issues between June 20, 2023 and February 12, 2026. The archive places it in contexts such as “Davidson artificially subtracts 3 OOMs to represent the lowest number”; “Davidson estimates the effective FLOP gap”; “Davidson just turns the whole problem into a parameter that he can plug into a standard economic model”. It most often appears alongside AGI, Ajeya Cotra, Bio Anchors.

Metadata

  • Category: People
  • Mention count: 2
  • Issue count: 2
  • First seen: June 20, 2023
  • Last seen: February 12, 2026

Appears In

Source Context

Recovered passages from the original issue text. When the raw archive preserved outbound links inside the source passage, they are listed directly under the quote.

Davidson On Takeoff Speeds
June 20, 2023 · Original source
I thought about this when reading What A Compute-Centric Framework Says About Takeoff Speeds, by Tom Davidson. Davidson tries to model what some people (including me) have previously called “slow AI takeoff”. He thinks this is a misnomer. Like skiing down the side of Mount Everest, progress in AI capabilities can be simultaneously gradual, continuous, fast, and terrifying. Specifically, he predicts it will take about three years to go from AIs that can do 20% of all human jobs (weighted by economic value) to AIs that can do 100%, with significantly superhuman AIs within a year after that.
Inline links: What A Compute-Centric Framework Says About Takeoff Speeds, this is a misnomer
As penance for my previous mistake, I’ll try to describe Davidson’s forecast in more depth.
You get what futurists call a “takeoff”. The first graph shows a world with no takeoff. Past AI progress doesn’t speed up future AI progress. The field moves forward at some constant rate. The second graph shows a world with a gradual “slow” takeoff. Early AIs (eg Codex) speed up AI progress a little. Intermediate AIs (eg an AI that can help predict optimal parameter values) might speed up AI research more. Later AIs (eg autonomous near-human level AIs) could do the vast majority of AI research work, speeding it up many times. We would expect the early stages of this process to take slightly less time than we would naively expect, and the latter stages to take much less time, since AIs are doing most of the work. The third graph shows a world with a sudden “fast” takeoff. Maybe there’s some single key insight that takes AIs from “mere brute-force pattern matchers” to “true intelligence”. Whenever you get this insight, AIs go from far-below-human-level to human-level or beyond, no gradual progress necessary. Before, I mentioned one reason Davidson doesn’t like these terms - “slow takeoff” can be fast. It’s actually worse than this; in some sense, a “slow takeoff” will necessarily be faster than a “fast takeoff” - if you superimpose the red and blue graphs above, the red line will be higher at every point1. CCF departs from this terminology in favor of trying to predict a particular length of takeoff in real time units. Specifically, it asks: how long will it take to go from the kind of early-to-intermediate AI that can automate 20% of jobs, to the truly-human-level AI that can automate 100% of jobs? (“Can automate” here means “is theoretically smart enough to automate” - actual automation will depend on companies fine-tuning it for specific tasks and providing it with the necessary machinery; for example, even a very smart AI can’t do plumbing until someone connects it to a robot body to do the dirty work. CCF will talk more about these kinds of considerations later.) In order to figure this out, it needs to figure out the interplay of a lot of different factors. I’m going to focus on the three I find most interesting: How much more compute does it take to train the AI that can automate 100% of the economy, compared to the one that can automate 20%?
Inline links: “takeoff”, Codex, 1
What Happened With Bio Anchors?
February 12, 2026 · Original source
In 2023, Tom Davidson published an updated version of Bio Anchors that added a term representing the possibility of recursive self-improvement. The new calculations shifted the median date of AGI from 2053 → 2043. This doesn’t explain why our own timeline seems to be going faster than Bio Anchors: even 2043 now feels on the late side, and anyway recursive self-improvement has barely begun to have effects.
Inline links: an updated version of Bio Anchors
But in 2025, John Croxton published a thorough report card on Davidson’s model. He took his numbers from Epoch, who used real data from the 2020 - 2025 period that earlier forecasters didn’t have access to, as well as the latest projections for what AI companies plan to do over the next few years. to with more formal projections. Most of his critiques apply to Bio Anchors too. We’ll be making use of them here.
Inline links: John Croxton published, from Epoch
Croxton found that Cotra and Davidson underestimated annual growth in effective compute: