Episode Details

AI ‘time horizons’ are mostly not about time (I think it's mostly ‘data’, but you’ll see where I’m unsure).

One chart from 2025 has become perhaps the most (in)famous in modern AI commentary.

For those in the know, ‘the METR graph’[1] is unusually compelling because it achieves what so few measures of AI progress have achieved: a somewhat meaningful Y axis (‘time horizon’[2]) as well as a somewhat predictable trend over time! (This is remarkably rare!)

Frustratingly, the only superficially available takeaway is something like, ‘the line goes up straight-ish over time’. This is better than nothing, but it's very dissatisfactory from the point of view of getting confidence in the predictions, because it exposes no deeper mechanism. This drives a lot of confusion and argument about the implications.

A deeper mechanism would be good for two reasons:

• It enables a sanity check on the trend, perhaps enabling more confidence in its predictions than we would sensibly allow with only the surface understanding.
• It gives some way to interrogate when and how the trend might change (because if the deeper mechanism gets deflected, the superficial projection would be broken, but a prediction based on the deeper mechanism might stay [...]

---

Outline:

(02:14) Attempting to find some mechanism in the METR graph

(02:19) Task 'length' and success modelling

(05:46) Relating hazard rate with frontier AI development

(06:56) Why does hazard rate shrink with date?

(10:27) Upshot

The original text contained 15 footnotes which were omitted from this narration.

---

First published:
May 24th, 2026

Source:
https://www.lesswrong.com/posts/zT76JcomKkdqo8tC6/a-slightly-mechanistic-theory-for-exponentially-increasing

---

Narrated by TYPE III AUDIO.

---

Images from the article:

---

---

Listen Now