Data up to 2022 became “pre-nuclear steel” for AI training

Artificial intelligence, intended to become the locomotive of technological progress, is beginning to slow down its own development. According to The Register, generative models have filled the internet with so much synthetic content that this creates a real technological dead end.

And we observe the formation of a “digital vicious circle”. New AI models are increasingly trained on materials created by previous AI versions. The result resembles the classic game of “telephone”. Where each subsequent participant passes on increasingly distorted information. In professional circles, this scenario received the name “model collapse”.

A striking analogy emerges with pre-nuclear steel! Just as metal produced before the first nuclear tests in July 1945 is valued for the absence of radionuclides, data up to 2022 becomes an invaluable resource due to minimal “contamination” by synthetic content.

Maurice Chiodo, research fellow at the Centre for Existential Risk Studies at Cambridge University, emphasizes: access to “clean data” is critically important for preventing model degradation.