AI Futures
Leading indicators
The scoreboard behind the weekly briefing — models, labor, safety, robotics, power, and policy signals.
Published by Steven Bryan
As of Oct 12, 2026
Refresh weekly
Alerts ALL CLEAR · C1 watch continues
Capability & models
| Indicator | Value | As of | Δ |
|---|
| AA Intelligence Index — top score | 58 | 2026-10-12 | 0 |
| AA top model | Claude Opus 5.5 (Sonnet 5.5 #2 at 56; Haiku 5.5 at 43) | 2026-10-12 | — |
| Arena Text top Elo | 1525 — Gemini 4 Argon (preliminary) | 2026-10-08 | 0 |
| Lab-reported long-horizon agent tasks | 12 hours (hold) | 2026-10 | 0 |
Labor & economy
| Indicator | Value | As of | Δ |
|---|
| US unemployment rate | 4.2% (hold) | 2026-09 | 0 |
| US nonfarm payrolls (month) | +29,000 (Jul+Aug revised −60,000) | 2026-09 | — |
| Youth (22–25) AI-exposed employment vs counterfactual | −19% (hold) | 2026-06 | 0 |
| Youth (22–25) high-exposure employment, year over year | −4.4% (35th straight month; hold) | 2026-08 | 0 |
| GS est. AI drag on US monthly payrolls | −16,000 / month (hold) | 2026-04 | 0 |
US–China & diffusion
| Indicator | Value | As of | Δ |
|---|
| Closed frontier lead (qualitative) | US lead, broad (Anthropic #1–2; OpenAI & Google 53) | 2026-10-12 | — |
| Open-weight ecosystem leader | China open stack (MiMo 46 / GLM 45 / Kimi 44) | 2026-10-12 | — |
Robotics & autonomy
| Indicator | Value | As of | Δ |
|---|
| Global humanoid shipments (2025) | ~13,300 (hold) | 2025 | 0 |
| Waymo weekly rides (reported) | ~500,000 (hold); $5B debt Oct 8 | 2026-10-12 | 0 |
| Tesla Optimus material factory usefulness | Not verified / early R&D | 2026-10 | — |
Safety, incidents, governance
| Indicator | Value | As of | Δ |
|---|
| Major agent containment incident (90 days) | Yes (carry) — Australia hearing Oct 6 is prior-incident follow-up; pause still on; no new breach this week | 2026-10 | — |
| Cited exec catastrophe odds band | 10–25% | 2026-10 | — |
| Governance readiness dial (from scorecard) | 31 | 2026-10-12 | 0 |
Energy & compute bottlenecks
| Indicator | Value | As of | Δ |
|---|
| Global data-center electricity | 485 TWh (hold) | 2025 | 0 |
| Energy/compute pressure dial | 56 | 2026-10-12 | 0 |
Markets (context)
| Indicator | Value | As of | Δ |
|---|
| Mag 7 concentration risk flag | Elevated | 2026-06 | — |
| Scenario weight: turbulent transition | 45% | 2026-10-12 | 0 |
What shipped (7 days)
| Indicator | Value | As of | Δ |
|---|
| Notable model/product releases | Claude Haiku 5.5 (Oct 7, AA 43) · Sonnet 5.5 cache-read cut · Waymo $5B debt (Oct 8) | 2026-10-12 | — |
| Notable API/price cuts | Haiku 5.5 $0.10/$0.50 (≤100k); Sonnet cache reads halved to $0.10 | 2026-10-12 | — |
Public trust, harm, adoption
| Indicator | Value | As of | Δ |
|---|
| Deepfake/election incident (7 days) | Brazil/US election deepfake backdrop continues — not catastrophic-scale | 2026-10-12 | — |
| Consumer adoption signal | none new primary MAU this week | 2026-10-12 | — |
| Enterprise adoption signal | McKinsey State of AI 2026 (~Aug vintage) | 2026-08 | — |
Regulation & geopolitics calendar
| Indicator | Value | As of | Δ |
|---|
| EU AI Act next milestone | General application ~2 Aug 2026 (high-risk later via Omnibus) | 2026-10 | — |
| US federal AI policy move (7 days) | No new statute; OpenAI Australia hearing Oct 6; FRONTIER Act still in committee | 2026-10-12 | — |
| Chip export-control change (7 days) | No new BIS rule | 2026-10-12 | — |
Price of intelligence
| Indicator | Value | As of | Δ |
|---|
| Frontier API list price (in/out per MTok) | Opus 5.5 $4/$20 · near-frontier $2/$10 · Haiku 5.5 $0.10/$0.50 (≤100k) | 2026-10-12 | — |
| Value-tier API price example | Haiku 5.5 $0.10/$0.50; Grok 4.7 ~$2/$6; Gemini 3.8 Flash ~$0.75/$3.75 | 2026-10-12 | — |
Needle movers
| Indicator | Value | As of | Δ |
|---|
| Abundance-up falsifier | Not triggered | 2026-10-12 | — |
| Doomsday-up falsifier | Not triggered for ≥5pp — C1 escalated watch continues | 2026-10-12 | — |