A Guardian letters exchange pushes back on the idea that openness is a defining feature of China’s AI development, while still crediting Chinese labs behind Qwen, DeepSeek and Kimi with helping popularize openly released models. The excerpt says that kind of competition benefits users broadly, especially where models can run on modest hardware in the developing world.
One response, from Prof Paul H Cleverley, argues that China’s openness claims deserve closer examination and uses GeoGPT, a geoscience system from Zhejiang Lab presented at the World AI Conference, as an example. According to the excerpt, GeoGPT is promoted to countries as open, yet under the model openness framework endorsed in a recent UN report, it would not qualify as open at all.
The excerpt says GeoGPT releases model weights but not training data or application source code, and that its governance committee answers to Zhejiang Lab itself. The broader exchange, which also includes Dr Claire Jenkins, frames the issue as larger than China alone, pointing toward the need for shared openness standards rather than national self-congratulation.