As the world's two leading AI powers, China and the United States cannot afford to let dialogue lapse even amid competition. While they have recently differed over model distillation, that dispute is not the core issue in their AI relationship.
Distillation is a well-established, general-purpose technique – "knowledge refinement," as Gai Keke of Zhongguancun Academy puts it. AI companies worldwide, Chinese and American alike, use it to optimize models and improve efficiency. Its essence is catching up, not surpassing; it creates no original breakthroughs out of thin air. Even OpenAI's Dean Ball says distillation is not the main driver of China's recent AI advances.
If distillation isn't the core issue, what is? Both Chinese and US experts agree: Sharing risks, jointly building rules and jointly governing should top the agenda – because when it comes to AI safety, danger knows no borders.