Author name: Hikmat Karimov, Rahid Zahid Alekberli

arXiv:2604.24076v1 Announce Type: cross
Abstract: As large language models (LLMs) are increasingly deployed in high-stakes and operational settings, evaluation strategies based solely on aggregate accuracy are often insucient to characterize system re…

arXiv:2604.24083v1 Announce Type: cross
Abstract: This study introduces the Kerimov-Alekberli model, a novel information-geometric framework that redefines AI safety by formally linking non-equilibrium thermodynamics to stochastic control for the ethi…