In 2015, DeepMind, owned by Alphabet, developed Artificial Intelligence (“AI”) AlphaGo to beat the reigning three-time European Champion, Fan Hui, in the board game “Go.” This victory marked a milestone in the development of AI because of the complexity of the game. Within the game, there are more possible board configurations than there are atoms in the known universe. This feat made AlphaGo’s success even more surprising to commentators who believed that because of Go’s complexity, a victory to this degree would not be possible for at least another decade. Two years later, AlphaGo was brought to China, the birthplace of Go, to challenge the world champion, Ke Jie. The face-off between AlphaGo and Ke Jie was described as “one of the greatest matches in history”— a five-hour battle of cognitive superiority. In the end, AlphaGo triumphed with a final score of 4-1.
Just as the steam engine represented the sunset of mankind’s dependence on human and animal power to accomplish physical tasks, AI like DeepMind represents the ability of robots to replace the need for human cognitive power. DeepMind is already able to predict acute kidney failure forty-eight hours earlier than human physicians. Motional is an AI capable of autonomous driving and is currently operating the “world’s most-established public robotaxi fleet.” IBM’s AI can read chest X-rays with accuracy and specificity that makes it statistically indistinguishable from modern radiology residents. In addition, ROSS can read through legislation, case law, and secondary sources to answer legal questions in casual and understandable language. However, a lawsuit by Thomson Reuters, alleging that ROSS stole content from Westlaw to build a competing legal research product, caused the startup to cease services. In short, properly trained AI can replace, and sometimes outperform, humans in strategic thinking.
However, AI can replace not only strategic thinking but also creative thinking. For instance, the painting Edmond de Belamy was generated by AI using generative adversarial networks sold at Christie’s Auction for $432,500. In addition, AI can write literary works. In China, a Shenzen court in Shenzhen Tencent Computer System Co., Ltd. v. Shanghai Yingxu Technology Co., Ltd. deemed that creative news articles written by Dreamwriter AI software are protectable under Chinese copyright law since “creative choices” were made by the software engineers who created the Dream writer software in the first place. This approach by the Chinese court is a noticeable departure from U.S. copyright law which requires the author of the work to be a human being. This policy was recently upheld in the 9th circuit case Naruto v. Slater, where the court determined that photographs taken by a macaque monkey of himself were not copyrightable on the grounds that the monkey “author” was not human.
Attitudes toward the copyrightability of AI appear to be similar to that of the early days of copyright protections to photographs. The United States initially viewed copyrights as merely chemical reactions which lacked the creativity and skill necessary to be afforded protection under the law. The Supreme Court did not recognize the copyrightability of photographs until its decision Burrow-Giles Lithographic Co. Sarony in 1884.
Since the Wake Forest Journal of Business & Intellectual Property last addressed the intersection between AI and intellectual property, the world has seen the rise of the next leap forward in the field. In addition to creative works, AI can now “invent.” DABUS is a creative AI that uses multiple disconnected neural nets. Each neural net consists of sub-neural nets which specialize in certain processes. For example, there is a neural net for combining letters into words, connecting puzzle pieces, and comparing two shapes. In a way, these sub-neural networks act as “memories” for DABUS, which can apply to new problems. This process is similar to how we have our own individual mental ideas and capabilities, and we can apply those ideas laterally to solve new problems. DABUS combines and detaches the individual nets to interconnect them into structures that can solve complex problems. This process is similar to one’s “stream of consciousness.” Thus far, DABUS has autonomously created two inventions: (1) a food container that uses fractal walls to provide rigidity, better grip, and better thermal conduction between touching containers, and (2) a signal beacon that uses a lacunar pulse train to trigger anomaly detection filters to attract enhanced attention.
Unfortunately for DABUS, the Patent and Trademark Office’s (“USPTO”) administrative interpretation of the patent statutes precludes an invention for being a “machine.” The USPTO relied on two fairly recent federal circuit cases to support its contention: Beech Aircraft Corp. v. Edo Corp. (where the federal circuit held that only “natural persons” can be inventors), and Univ. of Utah v. Max-Planck-Gesellschaft zur Forderung der Wissenschaften EV, (where the federal circuit held that a state could not be an inventor). The USPTO also noted that in 35 USC §101, Congress explicitly used the word “whoever,” which further supports USPTO statutory interpretation.
On September 2, 2021, the United States Districted Court for the Eastern District of Virginia granted summary judgment in favor of USPTO on a complaint brought under the Administrative Procedure Act (“APA”) alleging that the USPTO’s interpretation of the patent statutes was “arbitrary, capricious, [and] an abuse of discretion . . . ” under §706(2)(A). Under the APA, courts are highly deferential to agency interpretation; however, such interpretation is not necessary since the plain language of the patent act is unambiguous. The America Invents Act (“AIA”) defines “inventor” in §100(f) as “the individual … who invented or discovered the subject matter of the invention.” While the AIA is silent as to the meaning of “individual,” the district court interpreted it as referring to a “natural person.” This interpretation is supported by both the dictionary definition and by Congress’s clear reference to a natural person.
The status of AI under the law is no longer a theoretical question but a reality which the law might not be prepared for. While it is tempting to hope that AI will integrate neatly into existing rules, technologies like DABUS, Dreamwriter, ROSS, and DeepMind have shown that AI will almost certainly be this decade’s disruptive technology and will change the world just as personal computers and the internet did in the 1980s and 1990s. Judge Benjamin Cardozo once said, “new generations bring with them their new problems which call for new rules, to be patterned, indeed, after the rules of the past, and yet adapted to the needs and justice of another day and hour.” While the problems associated with AI have yet to become clear, it is unmistakable that AI is here, and the legal system needs to be ready to ensure that this technology ultimately improves people’s lives.
Benjamin Suslavich is a second-year law student at Wake Forest University School of Law. He is a certified Chemical Engineer and holds a Masters of Science in Metallurgical and Process Engineering as well as a Bachelors of Science in Materials Engineering from Montana Technological University.