Do LLMs Reason, or Do They Just Predict Math Text?

4 points by beagle3 60 days ago | 2 comments

gus_massa 60 days ago |

The best AI is predicting math text better than the me artesanily reasoning the solution. Thats good enough for me.

beagle3 60 days ago |

A critique of the KisMATH paper. Bottom line: Headline-claim inflation factor: roughly 3x–4x

gus_massa 60 days ago |

The best AI is predicting math text better than the me artesanily reasoning the solution. Thats good enough for me.

beagle3 60 days ago |

A critique of the KisMATH paper. Bottom line: Headline-claim inflation factor: roughly 3x–4x