What will be true of the SOTA AI on the FrontierMath benchmark, before 2026?
➕
Plus
6
Ṁ390
2026
57%
Transformer-based architecture
45%
Developed by Google Deepmind
45%
Narrow domain of knowledge. ie Does not know random facts such as when Google was founded, or who won the 1960 presidential election.
45%
Based on Symbolic AI (https://en.wikipedia.org/wiki/Symbolic_artificial_intelligence)
43%
Developed by OpenAI
39%
Part of the o1 family of models (o1, o2, etc. and variations)
34%
Over 1T parameters
34%
Developed by a non-British and non-American company
31%
Part of the GPT-N family of models (GPT-5, GPT-6, and variations)
27%
Part of the AlphaProof family of models (AlphaProof N and variations)
22%
Energy-based Model (https://en.wikipedia.org/wiki/Energy-based_model)

An option resolves YES if it is true about the AI model, or program, known to be State of the Art in terms of the FrontierMath benchmark, at the end of the year 2025. It resolves NO otherwise.

You're welcome to add any interesting facts that might or might not be true about the state of the art in math problems, as defined by achieving the highest score on the FrontierMath benchmarks.

I reserve the right to cancel any option that is too vague, too improbable, etc.

See also:
/Bayesian/what-will-true-of-the-sota-ai-on-th-y0LE5uE9n9 (This market)
/Bayesian/what-will-true-of-the-sota-ai-on-th-ROldIhZZgt
/Bayesian/what-will-true-of-the-sota-ai-on-th-RQptyR5uO8

/Bayesian/will-an-ai-achieve-85-performance-o-hyPtIE98qZ
/MatthewBarnett/will-an-ai-achieve-85-performance-o

/Bayesian/will-an-ai-achieve-30-performance-o

Get
Ṁ1,000
and
S3.00
© Manifold Markets, Inc.Terms + Mana-only TermsPrivacyRules