Math for AI safety: an invitation for mathematicians

Abstract:Artificial intelligence threatens to outrun human understanding and control. New mathematics is needed to design AI that is legible, steerable, and cooperative with humanity. I organize this invitation by mathematical field, so you can turn straight to your own: logic and game theory for cooperation; probability for agency and world-models; algebra and representation theory for learned features; analysis and geometry for generalization and training dynamics. Each section ends with an open problem that is accessible to a working mathematician with no prior experience in AI safety.

History and Overview (math.HO); Artificial Intelligence (cs.AI)

I.2.0; I.2.6; I.2.11; F.4.1; G.3; G.1.6; K.4.0

arXiv:2609.15289 [math.HO]

(or arXiv:2609.15289v1 [math.HO] for this version)

https://doi.org/10.48550/arXiv.2609.15289 Focus to learn more arXiv-issued DOI via DataCite (pending registration)

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.

Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.