By: Sydney Johnson
Can machines think? In 1950, British mathematician Alan Turing proposed this question and answered in the affirmative.[1] Recent advancements in artificial intelligence (“AI”) tend to support this answer. If we are to encourage further innovation of this technology, we must first answer an equally important question: is AI patentable?
From Siri to self-driving cars, AI seems like something straight out of a science fiction novel. In fact, AI technology has been growing so rapidly that we must evaluate whether our patent doctrine needs an update. As it stands, there are a number of problems with patenting AI, including subject matter eligibility and enablement. Will patent law bow down to this tech giant, or will a lack of available protections hinder innovation?
Patent law does not offer protection to ideas centered around laws of nature, natural phenomena, or abstract ideas.[2] In Bilski v. Kappos, the Supreme Court held that Bilski’s invention, a process for hedging against risks in purchasing energy commodities, was not patentable because it was based on an abstract idea.[3] If the goal of AI is to create a machine that “thinks” like the human brain, it is likely that the invention will be ineligible because it is based on abstract thought. The analysis set forth in Alice Corp. Pty. Ltd. v. CLS Bank Int’l proposes two questions: (1) whether the claim is directed to a patent-ineligible concept and (2) whether the claim includes an inventive concept.[4] Some artificial intelligence inventions may pass the Alice test, but it is currently unclear what courts will consider an inventive concept for this field of technology.[5]
More advanced forms of AI, “black box” AI, have opaque reasoning; therefore, it is unclear how the system reached its conclusions.[6] This may result in a problem. 35 U.S.C. § 112 tells us that an invention must be described so that a person having ordinary skill in the art (“PHOSITA”) can make and use the invention without undue experimentation. If the patent applicant cannot specify how the artificial intelligence analyzes data, how will the PHOSITA be enabled to make and use the invention?
In evaluating enablement issues with AI, it is helpful to take a look at the Wands factors: the breadth of the claims, the nature of the invention, the state of the prior art, the level of the PHOSITA, the level of predictability in the art, the amount of direction provided by the inventor, the existence of working examples, and the quantity of experimentation needed to make or use the invention based on the content of the disclosure.[7] Future case law regarding the Wands factors in relation to AI will provide us with a better picture of the enablement problem.[8]
If the United States hopes to surpass China in the race to sentient AI, these problems require immediate solutions. More specifically, solutions in line with the purpose of patent law: to encourage innovation.
[1] Cheston Tan, Artificial Intelligence: Will Computers Pass the Turing Test by 2029? Does it Matter?, Harvard University: SITN (July 3, 2012), http://sitn.hms.harvard.edu/flash/2012/ai/.
[2] 35 U.S.C. § 101.
[3] Bilski v. Kappos, 561 U.S. 593 (2010).
[4] Alice Corp. Pty. Ltd. v. CLS Bank International, 573 U.S. 208 (2014).
[5] Id.
[6] Ariel Bleicher, Demystifying the Black Box that is AI, Scientific American: Engineering (Aug. 9, 2017), https://www.scientificamerican.com/article/demystifying-the-black-box-that-is-ai/.
[7] In re Wands, Research & Diagnostic Systems, 665 F.3d 1269 (Fed. Cir. 2012).
[8] Id.