My response to the Chinese ambassador to the UK, published in The Guardian: he’s right that AI’s future depends on openness, but openness needs honest measurement.
Take GeoGPT, the hosted Large Language Model (LLM)-driven application system for the geosciences / geological sciences. It is developed by the Alibaba and Chinese Government founded Zhejiang Lab. GeoGPT was showcased last month at the World AI Conference in Shanghai, China’s flagship AI event. It is promoted as a technological “public good”, “fully open science” and “open-source”. However, under the Model Openness Framework (MOF), cited in a recent UN report, its conversational models would not even qualify as open models.
It releases only some of its model weights, none of which are released under an Open-source Initiative (OSI)-compliant licence. No actual AI training datasets have been released for GeoGPT (e.g., the 280,000 filtered open-access geoscience articles for continued pre-training nor the 89,000 Question and Answer pairs for supervised fine-tuning reportedly created using OpenAI’s ChatGPT). None of the application layer agentic source code has been released.
None of this is unique to China. The US private-enterprise model raises much the same concerns. That is really the ambassador’s own point extended: no country’s ecosystem is quite as open as its diplomacy suggests. If the UK is to gain from the cooperation he proposes, it should press for shared openness standards like the UN recommended Model Openness Framework as the basis for any joint work. Honest measurement, not just the language of openness, is where real partnership begins.
Link to full published letter: https://www.theguardian.com/technology/2026/aug/07/china-ai-ecosystem-is-not-as-open-as-it-claims-nor-is-any-other-country
#AI #OpenSource #AIPolicy #OpenScience #AI4Good #digitalpublicgood #China #sciencediplomacy #geology #geoscience #earthscience #EO
A Geodyssey - Geoscience Text Analytics and Enterprise Search Research 
Leave a comment