2 Comments

Meng, thank you for the thoughtful write-up. Do you happen to know why China leads the world in AI patent but lags far bhind the US and EU in producing high-quality AI models?

I heard that China filed 5x more pantents than the US in AI-area for 10 years leading to 2023. A Stanford research also says China accounts for ~60% of AI patents filed vs. 20% by the US.

You wrote, “Building an AI model is like making computer chips. It needs lots of money and research. It's not easy to do with just open-source code,“ but what are the differences between filing an AI ptent and developing an AI model?

Expand full comment

The protection of intellectual property (IP) in the field of AIGC (AI-Generated Content) urgently requires valuable recommendations and case studies across various stages, including application, rights protection, and even early R&D and model development.

- Patent Applications Embedding Non-Improvement Points: Refers to patent applications where the protected technical solution includes content generated by AIGC models as part of the non-improvement points.

- Patent Applications Embedding Improvement Points: Refers to patent applications where the protected technical solution includes AIGC model-generated content as part of the improvement points.

- Patents Protecting the Training Process: Refers to patent applications where the protected technical solution involves improvements related to the training process of the AIGC model.

- Patents Protecting Application Scenarios: Refers to patent applications where the protected technical solution involves improvements related to applying the AIGC model to a new scenario.

Expand full comment