Research updates

Live panel did not return all three estimates (xAI failed or gave an unusable answer). A round is only live when all three are valid, so this round is a placeholder at the previous score rather than an average of two.

OpenAI estimated 0.80. Particle-dominated dark matter remains plausible, but gravitational abundance evidence does not identify its constituents. The checked sources establish neither independently replicated particle detections with a… Anthropic estimated 0.60. Most cosmologists expect dark matter to be mostly particles beyond the Standard Model. That is why the probability sits above 0.5. The criterion needs independent non-gravitational detections of a common species, and… xAI estimated 0.64. PDG 2025 treats WIMPs, axions, and sterile neutrinos as unconfirmed candidates; the 3.5 keV feature is disfavored and direct searches give limits, not a calibrated abundance. No independent non-gravitational detections… The panel disagreed by 0.20, so read this average as soft. The three models did not return a unanimous confident binary, so the market is not decided.

OpenAI estimated 0.57. I interpret the target as fixed-vegetation equilibrium sensitivity, not effective sensitivity or AR6’s broader feedback convention. Evidence modestly favors exceeding 3 K, but the principal multi-line constraints span… Anthropic estimated 0.08. AR6 gives a best-estimate ECS of 3 K with a very likely range of 2–5 K. That interval crosses 3 K, so it fails the YES condition that the 95% interval lie wholly above 3 K. It also fails the NO condition that the upper… xAI estimated 0.50. No eligible multi-line assessment places the 95% interval for this equilibrium sensitivity wholly above 3 K, nor is the 95% upper bound at or below 3 K. IPCC AR6 and Sherwood et al. 2020 center near 3°C (about 2–5 K and… The panel disagreed by 0.49, so read this average as soft. The three models did not return a unanimous confident binary, so the market is not decided.

OpenAI estimated 0.35. I interpret the hypothesis as actual living ocean cells, not merely habitable conditions. Cassini measurements support available nutrients and organic chemistry, making life plausible, but the retrieved evidence… Anthropic estimated 0.02. No sample from Enceladus has shown cell-bounded metabolism and reproduction, and nothing has been independently reproduced. Cassini found a habitable-looking ocean, organics and H2, but these are not life detections.… xAI estimated 0.38. No peer-reviewed, independently reproduced observation of cell-bounded metabolism and serial reproduction from traceably ocean-derived Enceladus samples exists. Plume organics, methane and habitability data do not meet… The panel disagreed by 0.36, so read this average as soft. The three models did not return a unanimous confident binary, so the market is not decided.

OpenAI estimated 0.25. The retrieved calculations favor hcp, lowering my support for the proposed negative free-energy difference. However, the exact-state 2023 calculation lets bcc transform rather than measuring its constrained-basin free… Anthropic estimated 0.20. Published first-principles results disagree on the bcc–hcp ordering of iron near inner-core conditions. Belonoshko and co-workers favour bcc, while most free-energy studies and the named crystal-structure-prediction… xAI estimated 0.16. At the stated state point, the peer-reviewed T-USPEX/PBE calculation finds a bcc seed relaxes to hcp, so bcc is not stable in its crystalline basin and no constrained G_bcc-G_hcp 95% interval wholly below zero exists.… The three models did not return a unanimous confident binary, so the market is not decided.

Live panel did not return all three estimates (xAI failed or gave an unusable answer). A round is only live when all three are valid, so this round is a placeholder at the previous score rather than an average of two.

OpenAI estimated 0.50. I interpret the hypothesis as allowing arbitrarily small Majorana components, not requiring light-neutrino-dominated decay. Current searches provide limits rather than replicated signals, and no verified exact… Anthropic estimated 0.03. No 0νββ signal exists. The latest results are limits: KamLAND-Zen T1/2 > 3.8e26 yr, LEGEND-200 > 1.9e26 yr, and CUORE's limit. Neutrinos could still have a Majorana mass component, so the YES criterion is unmet. Nothing… xAI estimated 0.50. No experiment reports a neutrinoless double-beta signal at 5σ. LEGEND-200 combined with GERDA and MAJORANA finds no evidence, and PDG 2025 quotes only half-life limits (KamLAND-Zen). No result establishes an exact… The panel disagreed by 0.47, so read this average as soft. The three models did not return a unanimous confident binary, so the market is not decided.

OpenAI estimated 0.30. The strongest verified search found no signal and reported a 90% lower limit of 2.4×10^34 years, leaving lifetimes on either side of 10^35 years possible. PDG’s October 2025 review still cites that result, with no… Anthropic estimated 0.20. No experiment has seen a p→e+π0 signal, so the YES condition (a 95% interval wholly below 10^35 yr with a signal established) is not met. Current limits, about 2.4×10^34 yr from Super-Kamiokande, are well below 10^35… xAI estimated 0.15. PDG 2025 and Super-Kamiokande (Takenaka et al., PRD 2020) report no p→e+π0 candidates and a 90% CL lower limit τ/B > 2.4×10^34 yr, below 10^35 yr. That is neither an established decay signal with a 95% interval wholly… The panel disagreed by 0.15, so read this average as soft. The three models did not return a unanimous confident binary, so the market is not decided.

OpenAI estimated 0.95. I read the hypothesis as an unrelativized separation under the specified uniform approximation definition. Conditional sampling-hardness results strongly favor separation, but the checked primary reference requires… Anthropic estimated 0.02. I found no unconditional ZFC proof that SampBQP differs from SampBPP, and no proof that they are equal. An unconditional separation would likely imply unresolved results such as P≠PSPACE. The known results rely on… xAI estimated 0.58. No unconditional ZFC proof separates or equates SampBQP and SampBPP under the stated uniform total-variation definition. Aaronson–Arkhipov hardness is conditional on polynomial-hierarchy or average-case conjectures… The panel disagreed by 0.93, so read this average as soft. The three models did not return a unanimous confident binary, so the market is not decided.

OpenAI estimated 0.90. Rigorous finite verification supports the classical hypothesis, but cannot exclude an off-line zero at greater height. Clay still lists the problem as unresolved, and the checked sources provide neither an independently… Anthropic estimated 0.03. No accepted, independently checked proof or counterexample of the Riemann hypothesis exists. MathWorld, citing Clay (Sep 2026), lists it as unsolved. Numerical checks and heuristics favor truth, but they don't meet the… xAI estimated 0.86. The classical Riemann hypothesis remains an open Clay problem: no complete unconditional ZFC proof has been independently checked, and no certified zero with real part other than 1/2 is known. Verification of the first… The panel disagreed by 0.87, so read this average as soft. The three models did not return a unanimous confident binary, so the market is not decided.

OpenAI estimated 0.55. I interpret this as a claim of chemical capability, not that qualifying replication has already been demonstrated. Primary results support important components, but QT45 self-synthesis required a supplied hexamer and… Anthropic estimated 0.02. No evidence shows a ten-transfer, two-laboratory demonstration of replicase self-renewal from triplet substrates. QT45 copies itself with about 0.2% yield over 72 days. The freeze–thaw work replicates only a 29-nt… xAI estimated 0.28. Neither named primary result meets the ten-transfer, ≤10% retention, and concentration-recovery test. QT45 copies itself from trinucleotide substrates at ~0.2% yield in separate reactions; the freeze–thaw study shows… The panel disagreed by 0.53, so read this average as soft. The three models did not return a unanimous confident binary, so the market is not decided.

OpenAI estimated 0.76. The experimental critical-region signatures and molecular-model results support the hypothesis, interpreted as equilibrated bulk liquids rather than transient amorphous conversion. However, the accessible evidence does… Anthropic estimated 0.35. You et al. (Science 2026) is a peer-reviewed experiment that places an LLCP near 210 K and about 100 MPa, which is inside the domain. It is a single study using transient liquids heated from amorphous ices. It has no… xAI estimated 0.63. You et al. (Science, 2026) report critical-like scattering and a heat-capacity rise at 210±8 K after ultrafast heating of amorphous ices, and independent simulations place an LLCP near 198 K and ~127 MPa. That is… The panel disagreed by 0.41, so read this average as soft. The three models did not return a unanimous confident binary, so the market is not decided.

OpenAI estimated 0.10. The checked sources provide neither an unconditional ZFC proof of equality nor one of inequality; Clay continues to describe the problem as open. My low probability for equality is a tentative mathematical judgment, not… Anthropic estimated 0.01. P vs NP is still open. Clay lists no solution, and a 2026 status page says no accepted proof exists either way. The recent claimed refutation of ETH is a subexponential 3-SAT algorithm, which does not give a… xAI estimated 0.15. No accepted unconditional ZFC proof of P=NP or of P≠NP is in hand. Clay still lists the problem as open, and recent separation claims are either incomplete or unverified preprints. Absent a proof, failure to find a SAT… The three models did not return a unanimous confident binary, so the market is not decided.

Live panel did not return all three estimates (OpenAI, Anthropic, xAI failed or gave an unusable answer). A round is only live when all three are valid, so this round is a placeholder at the previous score rather than an average of two.

Live panel did not return all three estimates (OpenAI, Anthropic, xAI failed or gave an unusable answer). A round is only live when all three are valid, so this round is a placeholder at the previous score rather than an average of two.

Live panel did not return all three estimates (OpenAI, Anthropic, xAI failed or gave an unusable answer). A round is only live when all three are valid, so this round is a placeholder at the previous score rather than an average of two.

Live panel did not return all three estimates (OpenAI, Anthropic, xAI failed or gave an unusable answer). A round is only live when all three are valid, so this round is a placeholder at the previous score rather than an average of two.

Live panel did not return all three estimates (OpenAI, Anthropic, xAI failed or gave an unusable answer). A round is only live when all three are valid, so this round is a placeholder at the previous score rather than an average of two.

Live panel did not return all three estimates (OpenAI, Anthropic, xAI failed or gave an unusable answer). A round is only live when all three are valid, so this round is a placeholder at the previous score rather than an average of two.

Live panel did not return all three estimates (OpenAI, Anthropic, xAI failed or gave an unusable answer). A round is only live when all three are valid, so this round is a placeholder at the previous score rather than an average of two.

Live panel did not return all three estimates (OpenAI, Anthropic, xAI failed or gave an unusable answer). A round is only live when all three are valid, so this round is a placeholder at the previous score rather than an average of two.

Live panel did not return all three estimates (OpenAI, Anthropic, xAI failed or gave an unusable answer). A round is only live when all three are valid, so this round is a placeholder at the previous score rather than an average of two.

Live panel did not return all three estimates (OpenAI, Anthropic, xAI failed or gave an unusable answer). A round is only live when all three are valid, so this round is a placeholder at the previous score rather than an average of two.

Live panel did not return all three estimates (OpenAI, Anthropic, xAI failed or gave an unusable answer). A round is only live when all three are valid, so this round is a placeholder at the previous score rather than an average of two.

Live panel did not return all three estimates (OpenAI, Anthropic, xAI failed or gave an unusable answer). A round is only live when all three are valid, so this round is a placeholder at the previous score rather than an average of two.