LivetechWestern cycleAnchors pending

A publicly released AI model passes 50% on FrontierMath by end of 2027?

As of 1 Oct · 16:00 UTCResolves 31 Dec 20273 movers2 pressure moments
Timeline A · Threshold HitReality
78%
Crowd 55%Gap 0

By the end of 2027, a large language model available to the public — either from OpenAI, Anthropic, DeepSeek, or another lab — solves at least half of the FrontierMath benchmark problems correctly. This happens because scaling test-time compute (letting the model spend more processing power on each hard problem) keeps paying dividends, and the reasoning models that emerged in 2024–2025 continue improving along that path. The publicly accessible model sits on the Epoch AI leaderboard with a 50%+ score.

Timeline B · Still ShortReality
22%
Crowd 45%Gap 0

No publicly released model reaches 50% on FrontierMath by end of 2027, despite incremental progress. The benchmark's difficulty ceiling proves steeper than scaling reasoning alone can overcome — the problems may require novel mathematical insights or proof techniques that current model architectures cannot discover through compute scaling alone. The best public leaderboard entry stays stuck below the 50% mark, while the hardest problems remain unsolved.

crowd 55
The quiet paththe one nobody’s watching

A mathematical breakthrough in 2026–2027 renders FrontierMath training-attackable, collapsing the genuine-capability frame.

6%
The read
Reality still leans toward Threshold Hit.

Reality favors Threshold Hit — the pressures holding that outcome in place are stronger than the counter-pull. The main thing that could shift this is compute spending by the top three or four AI labs.

Where reality is heading

A, 80% bandCrowd today
Projected reality support to 31 Dec 2028A rises from 41% to about 52% with an 80% band of 33 to 71. B falls from 53% to about 44%. Quiet path near 4 to 6%. Crowd today: A 55%.100%7550250Crowd today · A 55%A ~52%B ~44%P ~4%NowJan 2027Jan 202831 Dec 2028
Reality crosses toward A in early 2028. Today’s crowd number sits at the top of A’s band, possible but optimistic. View as table

The movers, ranked by leverage

01
Rising-power eliteConsciousness-pattern template for an actor-class in a rising power whose decisions on technology, military doctrine, trade posture, and institutional strategy bend the trajectory of a rising-vs-ruling cycle. Operates in the B_leak boundary channel per AMENDMENT 47 §MCA-1 — what actors are reaching for.
leverage 78
Rising-power moves tend to accelerate rivalry trajectories.
02
Collective pull shiftMarket-class: discontinuous move in collective attention + capital toward an outcome, measurable in prediction markets (Polymarket / Kalshi / Manifold / Metaculus / PredictIt) and sector-rotation flows. Per AMENDMENT 47 §MCA-2, this IS the B_leak = consciousness measurement instrument where it covers an outcome. Direction contextual depending on which outcome the pull targets.
leverage 50
Prediction markets reprice when collective pull on an outcome moves.
03
Institutional trust decaySignal-class: measurable datum tracking collective trust in institutions — Pew/Gallup polling on Congress/Supreme-Court/media/science/business, approval ratings, cross-party cooperation rate, voter-turnout participation vs grievance, protest-event counts. Direction typically Timeline-B on democratic-haywire cycle.
leverage 55
Falling institutional trust accelerates the destabilization timeline.

What would change our mind

Signals toward A
  • Compute spending by the top three or four AI labs trajectory bending toward Threshold Hit
  • Breakthroughs in reasoning techniques that let models think longer at inference time without proportional cost per answer moving in support of Threshold Hit
  • Whether labs integrate formal math tools (like Lean) into training shifting in the direction that supports Threshold Hit
  • Export controls on advanced chips crossing the threshold toward Threshold Hit
  • A voluntary or regulatory pause on frontier model releases readings confirming the Threshold Hit direction
Signals toward B
  • Compute spending by the top three or four AI labs trajectory bending toward Still Short
  • Breakthroughs in reasoning techniques that let models think longer at inference time without proportional cost per answer moving in support of Still Short
  • Whether labs integrate formal math tools (like Lean) into training shifting in the direction that supports Still Short
  • Export controls on advanced chips holding against the shift toward Still Short
  • A voluntary or regulatory pause on frontier model releases readings confirming the Still Short direction

Pressure moments

US-China decoupling crystallizessemiconductor supply-chain split (chip-fab geography, EDA stack adoption) · telecom-equipment market split (Huawei + ZTE share of RoW markets) · payment-rail adoption (SWIFT vs CIPS)
3-7 year crystallization phase
High
AI capability discontinuityfrontier model benchmark saturation + new benchmark emergence · autonomous-agent task-completion rate on real-world browser/code benchmarks · model-collapse regression indicators
3-9 months at observable-step window; prelude years
High

Evidence