AI, global competition now hinges on distillation

Article by Mattia Dalla Costa for ItaliaOggi.

Global competition in AI is no longer driven solely by computing power, but increasingly by the ability to train efficient models while reducing costs, data requirements, and time. It is within this framework that knowledge distillation comes into play, at the heart of the dispute between OpenAI and DeepSeek, the latter accused of training smaller models by replicating the behaviour of GPT systems