MOSCOW, November 23. /TASS/. The content generated by neural networks cannot be currently checked for truthfulness, Director of the Institute of Systemic Programming Arutyun Avetisyan said.
"There is no technology with us to unambiguously determine whether it is a fake or not," Avetisyan said. "For example, the Kandinsky [a generating model created by Sber - TASS] generated certain content. [It would be] excellent [if we] have the technologies and we press a button and know exactly that it was generated. It turns out that no such technologies exist in nature," he added.
The director backed the idea of introducing compulsory marking for the content of generating neural networks. Countries "are bound" to implement artificial intelligence in all spheres in the modern world and "this cannot be stopped," he noted. This is why it is important to create trusted technologies, develop procedures for regulation of this sphere and hold public discussions of technologies.