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
WEEKLY SYNTHESIS24 Aug 2026

Weekly — 17–24 Aug 2026

AI for Science was the only ASI-core domain to move materially this week, driven by externally wet-lab-validated autonomous protein design and a dense cluster of AI-linked mathematical results. SynBio also advanced on positive Phase 3 personalized mRNA melanoma results. Intelligence and Robotics showed important capability signals but not enough independent reliability evidence to move maturity. ASI forecast remains unchanged.

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.

Weekly synthesis

ASI Readiness: 59.03/100 (+0.83 w/w). Singularity Readiness: 53.19/100 (+0.58). Central ASI estimate remains 2032–33; rolling probabilities remain 2%/6%/12%/30%/81% at 1/2/3/5/10 years.

What changed

AI for Science rose 56.4→59.0. The strongest evidence was autonomous Claude protein-binder design surviving independent synthesis and wet-lab measurement, plus a cluster of AI-led/co-developed open-problem results. Synthetic Biology rose 66.0→66.8 after Intismeran plus pembrolizumab met Phase 3 melanoma endpoints. Intelligence 63.7, Robotics 47.4, Compute 75.1, Energy 46.8, Materials 62.4, Longevity 36.3 and BCI 42.8 remain unchanged.

Key events

—Anthropic autonomous de novo binder design: 8.8/10, independently experimentally validated; binder design is not drug development.
—AI-led Yau–Tian–Donaldson cscK disproof preprint: 9.5/10 significance, full claim scope, independent mathematical verification pending.
—NVIDIA AVO: 8.8/10, seven-day autonomous kernel optimization plus 100 RHAE on ARC-AGI-3 public set; private-set/controlled-ablation evidence pending.
—Generalist GEN-1.5: 8.6/10, one-shot in-context physical learning; first-party, short-horizon and modest-success caveats keep Robotics flat.
—Intismeran personalized mRNA therapy: 8.2/10, positive Phase 3 RFS/DMFS topline; detailed effect sizes and OS pending.

Open Problem Closure Radar

33 records announced during the weekly window: 28 full-scope claims, 3 partial, 2 independently verified. 16 document some AI involvement; 10 are AI-assisted/co-developed/AI-led with significance >=7. The strongest unverified claim is the YTD cscK disproof. Erdős #501 is the clearest independently verified AI-linked closure this week via Lean plus external repository audit. Do not interpret full claim_scope as independent verification.

Bottlenecks

Long-horizon autonomous reliability remains first-order; progress nudged 42→43 after AVO but severity stays 78. Automated AI R&D remains 74 severity / 39 progress. Scientific validation worsened in relative terms: severity 58→60 while progress 50→51 because discovery volume outpaced independent checking; latest DVG is about 2.5.

Benchmark health

METR Time Horizon remains the preferred longitudinal autonomy trend but >16h estimates remain unreliable with the current task suite. HLE remains unsaturated. FrontierMath Tier 4 v2 is saturating and version breaks matter. DO Challenge reached 922/1000 but is now a saturating/public/harness-dependent benchmark. SWE-bench Verified remains retired and SWE-bench Pro degraded due task-quality/contamination concerns.