SINGULARITY COMPASS
OBSERVATORY DEGRADED

Semantic review is behind discovery. Evidence and Pulse may be incomplete until the backlog is cleared.

2 material first-party/frontier/open-problem candidate(s) have waited more than 6 hours for semantic review.

0 critical overdue · 2 material overdue
Last semantic import 19 Sept, 04:12
← Back to Pulse
DAILY REVIEW05 Sept 2026

Daily Pulse — 5 Sep 2026

AI-assisted mathematics dominates the recovered canonical evidence for this Daily: Claude completed a machine-checked Lean formalization of Fermat's Last Theorem, while separate preprints report material ChatGPT/Codex, GPT-5.6 and Fable 5 contributions to proof and counterexample work. Historical readiness values are shown on the current rebased methodology: ASI Readiness 59.03 (0.00), Singularity Readiness 53.19 (0.00), with the 5Y/10Y ASI forecast unchanged at 30%/81%.

ANALYSIS

Methodology note: Historical index values on this Pulse are normalized to the current methodology for comparability. The original as-published record remains unchanged in the canonical archive.

DAILY PULSE // 5 SEP 2026

ASI Readiness: 59.03 (0.00) Singularity Readiness: 53.19 (0.00) ASI forecast: 5Y 30% (0 pp) · 10Y 81% (0 pp) · central estimate 2032–33

Historical values are displayed on the current methodology. The September 1 decomposition reset is absorbed into prior history and is not a capability decline.

What changed

Claude completed a machine-checked Lean formalization of Fermat's Last Theorem. Anthropic reports an 11-day Claude Code-based multi-agent run. The final artifact is machine checked, but the complete autonomous workflow has not been independently reproduced by an external team. This formalizes known mathematics rather than solving a new open problem.

Sourcehttps://www.anthropic.com/research/formalizing-fermats-last-theorem

ChatGPT/Codex materially contributed to proofs in Cherednik's instanton-superpolynomial work. The revised preprint explicitly documents the AI contribution. The result is primary-source confirmed, with no external proof reproduction yet.

Sourcehttps://arxiv.org/abs/2607.25666

GPT-5.6 and Fable 5 assisted a tight n=4 correlation-gap proof. The author reports human correction/refinement plus computational verification of 2,745 Bernstein coefficient systems. The theorem remains an unreviewed preprint without independent reproduction.

Sourcehttps://arxiv.org/abs/2609.02659

ChatGPT 5.6 helped find a counterexample to the stable forking conjecture. Freitag and Mutchnik explicitly credit an interactive model search for the construction. One duplicate canonical entry for this same paper has been archived; this Pulse now points to the single surviving event.

Sourcehttps://arxiv.org/abs/2609.00436

Readiness and forecast

No change. The cluster strengthens evidence for AI participation in formalization, proof search and mathematical discovery, but external reproduction and cross-domain generalization remain insufficient for a calibrated Readiness or forecast move.

BOTTOM LINE: The signal is the clustering of concrete AI-assisted mathematical work, not a score jump. ASI Readiness remains 59.03 and Singularity Readiness 53.19 on the rebased scale.

_Historical correction (12 Sep 2026): methodology deltas were normalized and canonical duplicate references were cleaned while preserving the original publication timestamp._