Shafaqna English– A method that enables developers to compress large, high-performance AI models into more cost-effective and computationally efficient systems has emerged as the newest flashpoint in the escalating technological competition between the United States and China over supremacy in artificial intelligence.
This approach, commonly referred to as model distillation, involves taking the outputs generated by a powerful AI system and using them to train a smaller, more lightweight model. The resulting distilled model is capable of handling many of the same tasks as its larger counterpart, but requires significantly fewer computational resources to operate.
Although distillation has long been considered a routine and widely accepted technique within the AI research community, it has now become the focal point of a mounting legal and ethical debate. The central question is whether sophisticated AI capabilities can be replicated or transferred to third parties without obtaining explicit permission from the original developers and owners of the source models.