What is AI and how is it treated by the USPTO, EPO, and CNIPA?

In general, artificial intelligence (AI) is an automation of something that a human can do, or the simulation of intelligent human behavior by a machine. In other words, AI does what a human can do, but with much more data and processing of incoming information. Unfortunately, claiming AI in accordance with its typical definition is akin to asking for a Section 101 eligibility rejection of the subject in the United States. Europe and China have already updated their AI patent search processes. If the United States maintains its current machine intelligence research procedure in accordance with the abstract idea doctrine under the Alice and mayo Supreme Court, are we leaving this industry behind?

AI is an umbrella term that encompasses four main categories: reactive AI, memory-limited AI, theory of mind AI, and self-aware AI. Reactive AI includes machines that operate solely on the current data input; its decisions only take into account the current situation. Reactive AI does not draw conclusions based on the data entered. Examples of reactive AI include spam filters, Netflix show recommendations, and computer chess players.

Limited memory AI is able to make decisions based on data from the recent past and is able to improve its decision making processes over time. This is the category in which the vast majority of research and development and patenting takes place. Examples of AI with limited memory are autonomous vehicles that can interpret data from the environment and adapt automatic behavior if necessary.

The machines become more prescient in the next two categories – these include AI that can understand human emotions and make decisions based on that understanding in theory of mind AI. Even more futuristic is the self-aware AI: these machines are able to process the mental states of others and emotions, but also to have their own. Think of the robots in Wall-E or, darker, in I Robot.

When the original patent laws were drafted, legislators did not foresee that one day we would have machines with decision-making capabilities that mirror those of humans. As a result, the United States Patent and Trademark Office (USPTO), the European Patent Office (EPO), and the China National Intellectual Property Administration (CNIPA) all have eligibility restrictions for topics related to mental processes and patent duties that human beings can perform, especially what a human mind can do, including processing information and data and making decisions based on that information and data. The idea is to prevent the patent system from being abused in this way. But in view of the new technologies emerging in the field of AI, each of these offices has tried to update their research procedure to capture at best some part of this topic.

The CNIPA prohibits patenting methods of mental activity. Recently, the CNIPA has released Draft Examination Guidelines on examining inventions related to the improvement of artificial intelligence algorithms (such as deep learning, classification and clustering, and big data processing). In seeking a “technical solution” that could make machine intelligence patentable, the CNIPA suggests looking at improvements to algorithms and big data processing, whether the algorithms have a specific technical relationship to the internal structure of the computer system, and /or improvements to hardware computing efficiency or execution effect. The CNIPA considers improvements in data storage size, data transfer rate, and hardware processing speed as evidence of a technical solution required for patentability.

In March of this year, the EPO’s 2022 exam guidelines came into effect, stating explicitly: “[a] mathematical method can contribute to the technical character of an invention, i.e. contribute to the production of a technical effect serving a technical purpose, by application to a technological field and/or by adaptation to a specific technical implementation.” The EPO even goes so far as to state explicitly that mathematical formulas can be patentable when used in a specific technical implementation. Specific examples of technical effect improvements include efficient use of computer storage capacity or network bandwidth. EPO has published a series of examples of mathematical formulas that help demonstrate technical effect.

The USPTO issued its latest research guidance in 2019. The guideline placed a strong emphasis on technical improvements to a machine or the functioning of a machine for overcoming a subject that does not qualify for the rejection of abstract ideas. Technical improvements in the US specifically exclude end-user benefits, which is different from the new CNIPA and EPO practices, allowing user benefits to be a consideration of the technical effect. Also unique to the United States is our Supreme Court, which occasionally intervenes in patent cases, particularly with the Alice and mayo decisions, which replace any form of USPTO guidelines. The USPTO Guidance is constructed within the confines of the abstract idea/law of nature framework of: Alice and mayo, so it couldn’t go as far as what the CNIPA and the EPO guidelines have done by designating mathematical formulas as patentable when implemented by a machine, and to support big data processing and improvements to hardware processing speed. to be designated as patentable. So unfortunately we are a bit behind in terms of research procedure for machine processes and machine intelligence.

For more information, watch the latest episode of IP Practice Vlogs, available here.

Wen Xie image

Leave a Comment