InternationalUSRemember you can easily switch between MIP US and MIP International at any time

What makes AI subject matter patent eligible?




Youping Ma and Guoquan Yang of Bridgeon assess whether AI-related technology is patent eligible in China, analysing the way in which China interprets claims and providing advice on how to file an AI patent application in the jurisdiction

More and more products involve artificial intelligence (AI). AI technologies enhance the performance of products and make them more appealing to customers.

In order to protect such innovation and secure the market, patents may be one of the best approaches for a business. However, they may also be a source of frustration for an applicant, or even a patent attorney.

In the following article we will give practical guidance to people, who would like to protect their AI technology and business in China.

Is an AI-related technology patent-eligible in China?

A lot of people are confused by the patent application process in China.

Sometimes applicants or patent attorneys outside China are told that AI-related inventions are not eligible subject matter for a patent in China. However, we also find a lot of AI-related applications are granted.

When an applicant actually files his own AI-related application in China, he always expects to receive an official opinion from a patent examiner that the application does not relate to a statutory subject matter and is not patent eligible. Alternatively, the patent examiner will continuously challenge the inventive step or obviousness of the invention, even if he does not find any prior art which discloses the whole invention. The latter is actually another way for an examiner to imply that the invention is not patent eligible.

Some IP practitioners outside China feel that it is harder to get an AI patent in the US than in China. However, others have the opposite feeling.

Why is there so much confusion? Why is there so much chaos regarding AI patents? Why is it so difficult to give clear answers to an applicant about what he should do or what he should not do?

Firstly, an invention involving AI technology can be patent eligible in China as long as it is directed to a technical solution rather than an abstract idea and is properly handled, otherwise there would not be so many granted patents.

However, we have to admit that currently in China what subject matter is eligible and what subject matter is ineligible is still ambiguous for a great number of applicants and attorneys.

This is the same situation as in Europe and the US. In Europe, the criteria for eligibility is based on "technical character". In the US, several testing tools including the "machine or transformation test" have been proposed.

In China, eligible subject matter is generally determined based on technical problem, technical means and technical effect. This is very similar to the Europe testing tool of "technical character". Therefore, in practice, some examiners may directly use the "technical character" tool used in Europe.

The current approaches divide the issue into several sub-issues and seem rational and persuasive. However, when we look at the big picture, we find it actually fails to solve the problem and indeed makes it even worse. It complicates the problem. The original basic question of what is eligible subject matter is what the concept of "technical" is. It then becomes several sub-questions of technical problem, technical means and technical effect. But what problem is a "technical" problem? What means is a "technical" means? And, what effect is a "technical" effect? The current approaches still cannot answer the very basic question: what is the concept of "technical"?

If an applicant just follows the current guidance without help from an experienced patent attorney, he will get even more confused. He would not understand why his invention passed the eligibility issue and was granted or why his invention was rejected because of ineligible subject matter.

What kind of AI invention is eligible subject matter in China?

This question can be divided into two sub-questions. The first one is what kind of AI solution is eligible for a patent, and the second one is how to draft an eligible patent.

The first question can be rephrased in the following way: if drafted properly, what kind of AI invention is eligible?

In our world, there are natural substances or objects. There are also abstract ideas inside the heads of human beings. These two categories are not eligible for patents. However, if a solution combines the idea of a human being with a substance or object and can be separated and independent from the mental activity of a human being, it can count as eligible subject matter. A simple example of a rock hammer can be used to illustrate this concept. A rock is a natural object. A person's thought of making a striking tool is an abstract idea. The combination of this abstract idea of striking tool with the rock forms a technical solution.

The same criteria also applies to AI solutions. Actually, although it is called AI, it can still be separated from the human mind and work independently. In terms of technical solution, it is the same as other solutions.

We have applied this criteria in practice and helped many applicants to select suitable subject matter for patent applications. It has been shown that as long as the application is drafted properly, it works well in China.

The next question is how to draft a patent specification properly?

Understanding the way of interpreting a claim in China

Before the proper way to draft a patent is explained, the way of interpreting a claim must be clarified because this is the major reason why some patent attorneys or lawyers, especially US patent attorneys, feel confused about the examination of eligible subject matter in China. Sometimes, it seems hard in China, while at other times it seems relatively easy.

To help understand the manner of interpretation in China, we will compare it with the practice in the US.

In the US, according to the subject matter eligibility test of the Manual of Patent Examining Procedure (MPEP) 2106, the broadest reasonable interpretation (BRI) of the claim is established prior to examining a claim for eligibility.

However, in China, a claim will be interpreted from the standpoint of a person skilled in the art, or relevant art. Some judges further hold that the interpretation shall be made in compliance with the objective of the invention. In this regard, when interpreting a claim, the examiner will start from the viewpoint of a person skilled in the relevant art to ascertain whether he can find any technical means from his knowledge to implement what is said in the claim.

The difference between China and the US is that some non-statutory subject matter may be excluded through the interpretation of a claim in China. For example, for the US case of Mentor Graphics v. EVE-USA, Inc., carrier, which is non-eligible, was interpreted as covered by the claim. However, if it was interpreted in China, the claims might be interpreted as just including random-access memory because carrier waves cannot be covered from the reasonable perspective of a person skilled in the art.

US attorneys feel it is easier to argue eligible subject matter because the interpretation manner rather than the criteria for examination of eligibility is favourable to an applicant in China. On the other hand, they feel it is harder because the criteria for examination of eligibility is sometimes much stricter.

How to file an eligible AI patent in China?

Now, we know what inventions can be eligible. We will now look at a real application example to understand how the examiner thinks so that applications can be handled accordingly. Below is a claim of an AI-related application.

1. A method of generating combination features for machine learning samples, comprising:

  • acquiring a data record, including a plurality of attributes;
  • performing at least one binning operation for the plurality of attributes, to obtain bin group features composed of at least one bin feature, wherein each binning operation corresponds to a bin feature; and
  • generating combination features for machine learning samples by combining the bin group features.

This application was rejected by the examiner in China. The examiner held that this invention failed to specify the practical usage of the solution and failed to solve a practical technical problem by combining the solution with practical context. The examiner deemed that it was not a patent eligible subject matter.

What can we learn from this case?

We can see from this case that even if the claim does not describe the subject that performs the steps, the examiner does not challenge whether it covers an abstract idea. As explained above, the examiner supposes that in the context of a patent application, the steps are not performed in the mind of a person.

This is a different practice from that in US and is more favourable to the applicant. Of course, if the claim is drafted in the US manner, it is even better.

On the other hand, in the rejection decision, the examiner requires that the solution of the invention should be connected to a certain technical field to solve a practical technical problem.

This seems rational and is the criteria extensively accepted in China currently. The examiner would like the patent specification to clarify how the invention is incorporated into a certain technical scenario to achieve a technical effect.

However, which scenario is deemed "technical" and which effect is deemed "technical"?

The above invention example relates to features of original data for machine learning, which may be important and technical in the eyes of experts in the AI area, but are deemed non-technical in the eyes of the examiners currently. Accordingly, the effect thereof is also deemed non-technical in the eyes of the examiners.

Currently, patent examination lags behind technical development. Some time is still needed for examiners to fully understand that AI is not an abstract idea and is a technical solution.

In view of this, it is difficult for an applicant to challenge the thinking of the examiner, and what he should do is change his strategy. The case of AU Optronics (AUO) v Patent Reexamination Board (PRB) before the Beijing Higher Court may provide some help for an applicant to select a good way to handle an AI invention.

The AUO's application claimed a pixel array of contactless input panel, which included a plurality of pixels, each pixel including a sub-pixel and a positioning pixel on a side of the sub-pixel.

The PRB held that the claim merely provided a human-specified construction of a pixel array, which was not a technical means, could not solve a technical problem and could not achieve a technical effect.

The Beijing Higher Court re-investigated the technical problem to be solved by the invention and found that a user could input through the panel without contact. The court then held that this invention was to solve the technical problem of a contact panel which was easily damaged under pressure and was inefficient. Therefore, the Beijing Higher Court reversed the decision.

From this case, we can see that it will be much easier if an applicant can connect his invention with a known technical scenario. Besides, arguments or description of a patent specification based on known technical effects such as efficiency, speed, reliability are readily accepted under current practice.

AI heralds a promising new area. AI can be protected by patents and it is therefore worth filing an application. However, under the current practice in China, an applicant should carefully construct a strategy from the very beginning so as not to let the examiner connect his invention with ineligible subject matter.

Youping Ma
  Youping Ma is the founding partner of Bridgeon Law Firm and is a Chinese lawyer and patent attorney. With nearly 20 years of experience, she has extensive patent filing and prosecution expertise, and has worked on many large-scale IP infringement litigations and invalidations. She has advised many companies on issues such as IP strategy, overseas early warnings, patent pools and industry standards. She has a degree in computer science and law from Peking University. She is an executive council member of the Capital IP Services Association and Beijing Patent Attorneys Association, a member of the Standards Committee of the Capital IP Services Association, director of the Committee of the China Productivity Promotion Center and IP mentor for the Beijing Innovation and Entrepreneurship Incubator. She was awarded third prize for soft science innovation by SAPPRFT in 2017. She was also named a China Four-Star Patent Prosecutor in 2016 and a Beijing Outstanding Patent Attorney in 2014. In 2011, she was awarded first prize for soft science innovation by SAPPRFT. She is the author of the article Insights into the Design Patent System arising out of GUI Design Patent Invalidations in Patent Law Study 2017 compiled by the China National Intellectual Property Administration.

Guoquan Yang
  Guoquan Yang is a partner at Bridgeon Law Firm and is a Chinese lawyer and patent attorney. He has practised for 15 years and is experienced in patent filing and prosecution. He also excels in patent invalidation proceedings and litigation. He has represented some well-known companies such as Nokia, Sensormatic Electronics and China Telecom in landmark retrials before the China Supreme Court. He has expertise in R&D and IP disputes and has extensive knowledge of US and European IP practices. He is a trusted advisor to domestic and international clients alike. Mr Yang is a frequent speaker on intellectual property issues at conferences and forums. He is a member of the All-China Patent Agents Association (ACPAA) and a member of the AIPPI China Branch. He was named as a China Two-star Patent Prosecutor in 2015 and 2016. He is the author of the article Insights into the Design Patent System arising out of GUI Design Patent Invalidations in Patent Law Study 2017 compiled by the China National Intellectual Property Administration, and Navigating the minefield of priority rights in MIP's China IP Focus, 2018.


Comments






More from the Managing IP blog


null null null

null null null

null

September / October 2019

Should I stay or should I go? Private practice v in-house

In-house or private practice? It’s a question many lawyers and attorneys will probably ask themselves at some stage in their career. Max Walters speaks to those on either side of the fence about the benefits and drawbacks of both



Most read articles

Supplements