Development

A recurrent-depth trick hides more of a model's thinking

development2026-09-04

Claim

A reported recurrent-depth trick hides more of Astra's thinking, sparking speculation, a rebuttal that depth is within twice GPT-4's, one researcher's view that legible reasoning was always doomed, and worry that depth is simply a dial.

Reporting date: 2026-09-04 (not necessarily the event date)

Sources

Attribution and reporting linked by this record; links do not imply independent verification.

Welcome to September 4, 2026 (theinnermostloop.substack.com)

Knowledge graphopen in browser →

Reported in

Welcome to September 4, 2026

Actor

OpenAI

Evidence for

Supersedes

A lab president declares the AGI era and calls it not unreasonable

Referenced by

Cite or correct this record

Citation formatCitation text

A recurrent-depth trick hides more of a model's thinking
Innermost Loop KB — secondary record
Reporting date: 2026-09-04
KB permalink: https://nicholsn.github.io/innermost-loop-kb/developments/2026-09-04-legible-reasoning-was-always-doomed
Claim: A reported recurrent-depth trick hides more of Astra's thinking, sparking speculation, a rebuttal that depth is within twice GPT-4's, one researcher's view that legible reasoning was always doomed, and worry that depth is simply a dial.
Attributed sources:
- Welcome to September 4, 2026 — https://theinnermostloop.substack.com/p/welcome-to-september-4-2026
This citation does not imply independent verification of linked-source claims.

Suggest a correction — opens a GitHub draft for you to review and submit.