Artificial Intelligence (AI) has become a very hot topic in the field of patent law. As AI has achieved some significant successes and momentum in recent years, including among major high-tech companies, the implications of AI for patents have grabbed everyone’s attention. Both high-tech companies and law firms are trying to figure out what strategies to use to patent artificial intelligence and to patent the solutions that AI generates. It’s the Wild West all over again–everyone is scrambling to figure out the implications of AI for patent law, with no one providing clear solutions.

I wrote the book on patents and AI almost ten years ago: The Genie In The Machine: How Computer-Automated Invention is Revolutionizing Law and Business (Stanford 2009). This was the first, and perhaps still the only, book to address the impact of AI on patent law and strategy. In the book I both explained the challenges that AI poses for patent law and proposed concrete legal strategies that lawyers and high tech companies could use to maximize their legal protection relative to their competitors in the face of rapidly developing AI. I use these strategies for the benefit of my clients at Blueshift IP.

When I was writing the book and through to today, I think people are focusing on the wrong issues. When most people first think about the impact of AI on patent law, they tend to focus on the following red herrings:

• Will AI replace human inventors and make them obsolete?
• Should inventions generated by AI be patentable?
• Should a computer be capable of being listed as an inventor on a patent?
• Who owns the rights to an invention that’s created by AI?

I think these questions are either easy to answer under existing law or raise issues that are not of practical importance today in light of what AI can do right now. There are other, much more important issues, that AI is raising for patent law right now, but that are not receiving attention.

In particular, AI raises very challenging problems for how to apply the law of “nonobviousness” (also referred to as “inventive step”) in patent law. Although nonobviousness is notoriously hard to define and understand, imagine that you invent a new kind of chair and that what is new about it is that it has five legs arranged in some unusual configuration that makes the chair more stable than previous chairs. To obtain a patent on this chair, many requirements need to be satisfied. Two of them are what are called “novelty” and “nonobviousness.” Novelty means that no one can have ever have sold, patented, or published a description of your chair before. Even if your chair is novel, it isn’t necessarily patentable, because it must also not be obvious. A simple (although not entirely legally accurate) way to understand the requirement of nonobviousness is that it requires that your chair be a significant improvement over what people who invent chairs would think of creating, based on their knowledge of existing chairs and their skill at inventing chairs.

As this example illustrates, embedded within the concept of nonobviousness is an understanding of what people who normally invent chairs have the skill to invent. We call this “the skill of the person having ordinary skill in the art (field) to which the invention pertains.” To determine whether your chair is obvious, we ask whether the hypothetical “person having ordinary skill in the art” (PHOSITA) would have found it obvious to invent your chair. And if they would have, then your chair is obvious and not patentable.

To make this decision about whether your invention is obvious, the patent examiner has to draw a conclusion about how much and what type of skill the PHOSITA in the art of chair making has. The claim that I make is that as AI improves, and to the extent that inventors can use improved AI to augment their inventive ability by doing things such as automatically generating designs for potential inventions, the general level of skill of PHOSITAs in fields where AI is commonly used in the process of inventing will increase. In short, as AI effectively increases the inventive abilities of inventors, the level of skill of PHOSITA will increase.

The legal implication of the level of skill of PHOSITA increasing in any particular field is that it will, or at least should, become more difficult to prove that a purported invention in that field is nonobvious and therefore patentable. In short, improvements in AI will make it harder to obtain patents, to the extent that such AI is used by inventors to effectively boost their inventive skill. AI will raise the bar for obtaining patents, indirectly through the requirement of nonobviousness.

The reason I think that people haven’t seen this and instead ask a more general, blunt kind of question, such as, “Will AI eliminate the need for inventors?,” is that they don’t understand the subtle way in which improvements in technology impact how patent law are applied. This effect of improving AI on nonobviousness is not an all or nothing effect. It’s not going to make potential inventions go from patentable to unpatentable overnight. Instead, the effect will be gradual and subtle, will vary from field to field, and will need to be evaluated on a case by case basis by high-tech companies, patent attorneys, patent offices, and courts.

This impact of AI on patent law, although subtle, is significant and is an example of several ways in which the impact of AI sneaks into patent law in unexpected ways that are easy to overlook. But once you are aware of these effect, it is possible to develop and execute strategies to maximize your likelihood of obtaining a patent in the face of a heightened standard for nonobviousness, or–if you are a defendant in a patent lawsuit–strategies to maximize your likelihood of invalidating a patent for being obvious. I discuss this and many other implications of AI for patent law in The Genie in the Machine.

This is an example of what we do for our clients at Blueshift IP. We think ahead of the curve, using our expertise in computer technology and patent law to apply unique strategies for patenting developments in AI. If you are developing new AI technology or are using AI to develop new inventions, contact us to find out how we can help you maximize protection for your innovations.