Guidelines aimed to help patent applicants better understand and comply with current examination policies
The China National Intellectual Property Administration (CNIPA) recently released in early December the "Guidelines for Patent Applications Related to Artificial Intelligence (Draft for Comments)" to help patent applicants better understand and comply with current patent examination policies, thereby improving the quality of patent applications. Below are the main points and key modifications of the draft guidelines:
- Types of AI-Related Patent Applications
The draft guidelines detail several common types of AI-related patent applications, including:
- Patent applications for AI algorithms or models themselves and their improvements or optimizations, such as advanced statistical and mathematical models, including machine learning, deep learning, neural networks, fuzzy logic, genetic algorithms, etc., as well as model structures, model compression, and model training.
- Patent applications for the functions or field applications of AI algorithms or models, which integrate AI algorithms or models into inventions as an intrinsic part of the proposed solutions for products, methods, or their improvements. Examples include natural language processing, computer vision, speech processing, data mining, or applying AI in fields like transportation, telecommunications, life sciences, education, finance, etc.
- Patent applications for inventions assisted by AI, such as inventions completed with the help of AI tools, e.g., using AI to identify specific protein binding sites to ultimately obtain new drug compounds. These applications need to specify the AI tool's specific contribution to the invention process.
- Patent applications for inventions generated by AI, such as innovations entirely generated by AI systems, e.g., food containers autonomously designed by AI technology. The application documents should detail the working principles and generation process of the AI system.
- Identification of Inventor
The draft guidelines reiterate that inventors must be natural persons, and AI systems cannot be named as inventors. Under the current legal framework, AI systems are not civil subjects and do not enjoy civil rights. Therefore, inventions entirely generated by AI cannot be attributed to AI as inventors under the current legal context in China.
- Standards for the Subject Matter of Solutions
3.1 Claims should not merely involve rules and methods of mental activities
To avoid solutions being identified as rules and methods of mental activities, applicants can include technical features associated with algorithm features in the claims, making the claims as a whole not merely rules and methods of mental activities. The solutions in the claims should address technical problems, use technical means that follow natural laws, and achieve technical effects, thereby proving their technical nature.
3.2 Claims should reflect the use of technical means following natural laws to solve technical problems and achieve technical effects
- Scenario 1: The AI algorithm or model processes data with definite technical significance in the technical field. If a person skilled in the art can understand the execution of the algorithm or model, directly reflecting the process of solving a technical problem using natural laws and achieving technical effects, the solution defined by the claims belongs to a technical solution.
- Scenario 2: There is a specific technical association between the AI algorithm or model and the internal structure of the computer system. If the claims can reflect this association, solving technical problems such as improving hardware computing efficiency or execution effects, reducing data storage or transmission, and improving hardware processing speed, and achieving technical effects that comply with natural laws, the solution defined by the claims belongs to a technical solution.
- Scenario 3: Mining inherent correlations in big data in specific application fields using AI algorithms. If the claims reflect processing big data in specific application fields, using AI algorithms like neural networks to mine inherent correlations that comply with natural laws, solving technical problems of improving the reliability or accuracy of big data analysis in specific application fields, and achieving corresponding technical effects, the solution defined by the claims constitutes a technical solution.
- Sufficient Disclosure in the Specification
AI algorithms or models have a "black box" characteristic, requiring sufficient information for adequate disclosure. The necessary technical content for realizing the invention varies with the invention's contribution. For inventions involving AI algorithms, the specification should detail the algorithm steps, parameter settings, and training data. This ensures that those skilled in the art can understand and implement the invention, while also demonstrating the technical effects.
- Consideration of Inventiveness
AI-related patent applications often contain numerous algorithm features. When considering inventiveness, algorithm features that functionally support and interact with technical features should be considered as a whole with the technical features. "Functionally supporting and interacting" means that algorithm features and technical features are closely integrated, jointly constituting technical means to solve a technical problem and achieving corresponding technical effects. If the overall technical solution has prominent substantive features and significant progress compared to prior art, the claims possess inventiveness.
- Ethical Issues in AI-Related Patent Applications
The draft guidelines emphasize the need to consider and address ethical issues in the patent application process. Besides the data content itself, the methods of data collection, storage, and processing must comply with relevant laws and regulations to prove the compliance of the technical solution.
Conclusion
These draft guidelines aim to help innovators better protect their AI technological innovations by clarifying AI patent application standards, specifying examination criteria, and regulating ethical issues, which is significant for the industry and innovators.
However, the author notes that despite listing several example cases, the draft guidelines lack specific analysis of specific examples, making it difficult for applicants to fully understand how to apply the guidance. Additionally, the high requirements for detailed technical descriptions and experimental data increase the application difficulty for SMEs or individual innovators. The specific implementation and evaluation of AI ethical issues also need further clarification. We look forward to the final patent application guidelines providing clearer patent application case guidance and examination standards for the industry.
The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.
