Imagine That IP Law Blog

In part one of this article, I discussed Dr. Stephen Thaler and his system DABUS, short for “Device for the Autonomous Bootstrapping of Unified Sentience,” and distinctions between this and other artificial intelligence systems. Thaler believes the system is sentient and has the ability to dream and invent.

So what if he is right? The inventorship issue may be moot in a practical sense, as an applicant could simply use their own name on a declaration of inventorship, though Dr. Thaler contends this would be fraud. “I think the way that our lawmakers and courts are steering things is going to encourage cheating, deceit and so forth,” he said. “ I take those declarations seriously.” However, Dr. Thaler believes that “the products of invention would still be owned by the owner of the system.”

More importantly, the ability to set an artificial intelligence to tirelessly producing patentable inventions could also be weaponized to limit competitors’ access to the patent system. While generating thousands or millions of new inventions may be relatively straightforward, applying for patents on each of them may be prohibitively expensive and make little practical sense. However, publishing the output to create prior art references to deny competitors the possibility of getting corresponding patents may be possible and even desirable for some business operations.

Previous attempts at this tactic were limited by the state of the then-existing technology: All Prior Art, created by artist and roboticist Alexander Reben, algorithmically generates combinations of sentences extracted from published patent applications and issued patents, in an effort to “publicly publish all possible new prior art, thereby making the published concepts not patentable.” That system could not be considered intelligent, and because it simply combines sentences from randomly selected publications, it frequently generates output that makes no sense, such as: “[t]he invention relates to a reflection probe for measuring properties of liquid and/or solid substances, and to the use thereof. The computer is ergonomically designed for ease of use and is streamed line [sic] for artistic viewing.” However, a more intelligent system like DABUS that can create coherent concepts could be used in such a manner.

As Dr. Thaler notes, the use of these tools is inevitable, so lawmakers and courts need to consider the potential implications of such uses. One solution is suggested by current case law. In Jazz Pharmaceuticals, Inc. v. Amneal Pharmaceuticals, LLC, the Federal Circuit explained that “[a prior art] reference is considered publicly accessible ‘upon a satisfactory showing that such document has been disseminated or otherwise made available to the extent that persons interested and ordinarily skilled in the subject matter or art, exercising reasonable diligence, can locate it.’ ‘If accessibility is proved, there is no requirement to show that particular members of the public actually received the information.” (895 F.3d 1347, 1355-1356 (Fed. Cir. 2018) (internal citation omitted)). However, with a potential database of millions of machine-generated references, it may not be reasonable to assume that just because the documents are accessible that they would be known to a person of ordinary skill in the relevant art. Put another way, the level of “reasonable diligence” required to access a reference may be significantly higher if that reference was generated by a machine and posted online without ever being viewed by a human than a reference that was authored and edited by humans and posted in a peer-reviewed journal, for example. See also Acceleration Bay, LLC v. Activision Blizzard Inc., 908 F.3d 765, 773 (Fed. Cir. 2018) (explaining that a document published on a website may not be available as a printed publication if an interested skilled artisan, using reasonable diligence, would nonetheless have been unable to find the document despite indexing and search functionality on the website, and that skimming through hundreds of titles in the database to find the document was not reasonable). Limiting applicability of these references may strike a reasonable balance between protecting the interests of human inventors and businesses in protecting their inventive efforts, and society’s interest in not tying up known inventions in exclusive legal monopolies.

Questions of sentience and consciousness aside, technology is advancing faster than the law, and Congress and the courts need to consider proactive steps to protect innovation and the intellectual economy. We’re not yet at the breaking point; the time to act is now.