Development

A rival credits scale, long-horizon RL and a constitution

development2026-07-05

Claim

Google's Andy Coenen credited Fable's strength to three ingredients — sheer scale, relentless long-horizon agentic reinforcement learning, and a constitution that turns thousands of decisions into one legible compass — while practitioners reported that letting the model exercise judgement beats dictating rules and conserves tokens.

Reporting date: 2026-07-05 (not necessarily the event date)

Sources

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

Welcome to July 5, 2026 (theinnermostloop.substack.com)

Knowledge graphopen in browser →

Reported in

Welcome to July 5, 2026

Actor

Evidence for

Supersedes

Frontier labs hire philosophers to write AI constitutions

Referenced by

Welcome to July 5, 2026

Cite or correct this record

Citation formatCitation text

A rival credits scale, long-horizon RL and a constitution
Innermost Loop KB — secondary record
Reporting date: 2026-07-05
KB permalink: https://nicholsn.github.io/innermost-loop-kb/developments/2026-07-05-a-constitution-as-a-legible-compass
Claim: Google's Andy Coenen credited Fable's strength to three ingredients — sheer scale, relentless long-horizon agentic reinforcement learning, and a constitution that turns thousands of decisions into one legible compass — while practitioners reported that letting the model exercise judgement beats dictating rules and conserves tokens.
Attributed sources:
- Welcome to July 5, 2026 — https://theinnermostloop.substack.com/p/welcome-to-july-5-2026
This citation does not imply independent verification of linked-source claims.

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