[Linkpost] “A deep critique of AI 2027’s bad timeline models” by titotal

EA Forum Podcast (Curated & popular)

Content provided by EA Forum Team. All podcast content including episodes, graphics, and podcast descriptions are uploaded and provided directly by EA Forum Team or their podcast platform partner. If you believe someone is using your copyrighted work without your permission, you can follow the process outlined here https://ppacc.player.fm/legal.

8d ago 1:19:43

MP3•Episode home

This is a link post.

Thank you to Arepo and Eli Lifland for looking over this article for errors.

I am sorry that this article is so long. Every time I thought I was done with it I ran into more issues with the model, and I wanted to be as thorough as I could. I’m not going to blame anyone for skimming parts of this article.

Note that the majority of this article was written before Eli's updated model was released (the site was updated june 8th). His new model improves on some of my objections, but the majority still stand.

Introduction:

AI 2027 is an article written by the “AI futures team”. The primary piece is a short story penned by Scott Alexander, depicting a month by month scenario of a near-future where AI becomes superintelligent in 2027,proceeding to automate the entire economy in only [...]

---

Outline:

(00:45) Introduction:

(05:27) Part 1: Time horizons extension model

(05:33) Overview of their forecast

(10:23) The exponential curve

(13:25) The superexponential curve

(20:20) Conceptual reasons:

(28:38) Intermediate speedups

(36:00) Have AI 2027 been sending out a false graph?

(41:50) Some skepticism about projection

(46:13) Part 2: Benchmarks and gaps and beyond

(46:19) The benchmark part of benchmark and gaps:

(52:53) The time horizon part of the model

(58:02) The gap model

(01:00:58) What about Eli's recent update?

(01:05:19) Six stories that fit the data

(01:10:46) Conclusion

The original text contained 11 footnotes which were omitted from this narration.

---

First published:
June 19th, 2025

Source:
https://forum.effectivealtruism.org/posts/KgejNns3ojrvCfFbi/a-deep-critique-of-ai-2027-s-bad-timeline-models

Linkpost URL:
https://titotal.substack.com/p/a-deep-critique-of-ai-2027s-bad-timeline

---

Narrated by TYPE III AUDIO.

---

Images from the article:

Mathematical formula showing calculation of 'n' using logarithms and variables.

Chemical formula showing H equals H₂O, stylized with blue shading.

Mathematical equation showing H(minutes) equals 15 times a decay function.

Mathematical equation showing relationship between H, H₀, alpha, t, and T₀.

Mathematical equation showing change in momentum equals force times change in time

Mathematical equation showing time function t(n) with period notation

Mathematical equation showing time function t(n) equals sum from n=0

Line graph comparing original and adjusted data for internal release gap.

Graph showing AI model task completion times versus release dates, 2019-2027, featuring multiple models.

Mathematical equation showing V total equals V avg plus 1 divided by 2

Mathematical equation showing algebraic terminal velocity calculation formula.

Graph showing model curves and speedups data, titled

Mathematical formula showing entropy (H) equals H zero two to T

Graph showing AI coding task completion times from 2021-2028, titled

Timeline diagram showing AI progress gaps from Re-Bench Saturation to Superhuman Coder.

Complex mathematical equation with H and H₀ variables and exponential terms.

Mathematical equation showing H equals H subscript 0 plus time function

Graph showing time horizon projections with exponential and hyperexponential curves through 2030.

Text excerpt discussing METR suite's LOC scaling and doubling time calculations.

Statistical distribution graph showing predicted date for crossing score 1.5, centered around 2026-2027.

Graph showing METR data with exponential and superexponential trend lines through 2030.

Graph showing time horizon projections with six different curve models, 2019-2032.

Apple Podcasts and Spotify do not show images in the episode description. Try Pocket Casts, or another podcast app.

256 episodes

#Tech #Society #Philosophy #EA Forum Team #Technology #Non-Profit #Future