Filed with the Supreme Court on April 18, 2023. Amycus We encourage judges to rule on the question of whether artificial intelligence (AI) is appropriate to be the inventor of a patent application. The issue arose in a certiorari writ petition recently filed by Dr. Stephen Thaler.
Thaler claims that his DABUS AI is the sole inventor of the two ideas he has filed for patents on, and so far there is no dispute. The USPTO declined to examine these inventions because DABUS was listed as the inventor. The USPTO has argued that inventorship is limited to natural persons only. This view also persuaded a district court judge and was confirmed by the Federal Circuit. Dr. Thaler’s petition to the Supreme Court followed.
Although patent law does not explicitly consider non-human inventors, it does not rule out the possibility either. Given that the constitutional principle behind that law is to “promote the progress of science and the useful arts” and to encourage innovation, the term “inventor” can be used to define a broad and non-limiting In fact, the Supreme Court, in considering the types of patentable inventions, recognized that “times change” and that “technology and other innovations advance in unexpected ways.” recognizing.
A key principle behind our interest in this case is the rapid expansion of generative AI. Generative AI can produce new creative works of writing, art, and music, not just inventions. This technology, which only became widely available to the public last year, is fundamentally different from traditional AI models that perform classification, clustering, and prediction. As pointed out in the brief:
The move from old traditional AI techniques to generative AI is unlike a calculator that can now perform multiplication between even larger numbers. Instead, this analogy calculator can infer and derive new useful formulas and properties from first principles. In short, the capabilities of generative AI are: Sui Generith— Fundamentally different from traditional AI and traditional computing.
Moreover, it is widely expected that the capabilities of generative AI will continue to grow as models become larger and finer-tuned. So if the courts don’t address this issue now, the legal system will have to deal with it in a piecemeal fashion in the years to come.
Because AI-generated inventions are currently unpatentable under the USPTO and lower court DABUS decisions, courts implicitly encourage entities to conceal AI-generated inventions. You must intervene to prevent When patents are not available, these companies look to protect their trade secrets, even inventions that have the potential to bring great benefits to mankind (e.g., new and useful products, including life-saving drugs). The willingness to commercialize such inventions may decline. This puts the United States at a competitive disadvantage compared to countries with more forward-looking policies. Additionally, lack of disclosure can lead to a lack of regulatory oversight on how generative AI models are trained, operated and used. This could help encourage proprietary practices regarding AI-generated inventions and limit their use to only a few players.
Finally, we address ancillary issues regarding ownership of AI inventions. Concerns in this area are easily addressed by existing agency and contract law. We don’t see much difference between the AI-as-inventor scenario and the human-as-inventor scenario. In both cases, the inventor and owner (assignee) of the patent application may (and usually are) different. Similar contractual arrangements can handle situations where ownership of an invention is divided between different parties, such as the entity that trained the AI, deployed the AI, or inspired the AI to invent it.
Considering all of this, I don’t see any compelling reason to exclude AI as an inventor in patent applications, and we recognize that there are many benefits to acknowledging AI as an inventor.Therefore, we, together with others Amiciasked the court to take the case and rule in favor of Dr. Thaler.
*MBHB Law Clerk Dr. Yuri Levin-Schwartz also contributed significantly to the overview.
[View source.]
