Generative AI & Law: Title 35 in 2024++ with Non-human Inventors

The first patent ever filed and granted in the United States was issued on July 31, 1790, to Samuel Hopkins. This patent was for a new method of making potash, an ingredient used in fertilizer.

The visual shows Samuel Hopkins working on the new method of making potash, an ingredient used in fertilizer, which became the first patent ever filed and granted in the United States, on July 31, 1790.

The patent was signed by President George Washington, Secretary of State Thomas Jefferson, and Attorney General Edmund Randolph, reflecting the high importance the new nation placed on the protection of inventions. This event marked the beginning of the formal patent system in the United States, a system that has since played a crucial role in fostering innovation and economic growth.

Title 35 of the United States Code

Title 35 of the United States Code (U.S.C) governs patents in the United States. It outlines the rules and procedures for obtaining a patent, including the requirements for patentability such as novelty, non-obviousness, and usefulness. The code also details the rights and obligations of patent holders, which include the right to exclude others from making, using, or selling the patented invention for a limited period. Additionally, it addresses matters like patent infringement, remedies for infringement, and the structure and function of the USPTO. This code is fundamental in shaping U.S. patent law and protecting intellectual property rights.

Surroundings of Samuel Hopkins' workshop as it would have appeared in 1790.

Inventors as Natural Persons

U.S.C Title 35, which governs patent law in the U.S., emphasizes that inventors must be natural persons. This is evident in its language; for instance, Section 101 uses the term "Whoever invents or discovers…" implying a natural person. Furthermore, Section 115 explicitly uses terms like "himself," "herself," "individual," and "person."

Image depicts the Patent Office in 1802, with the appointment of its first full-time employee, Dr. William Thornton

The United States Patent and Trademark Office (USPTO) interprets these references as excluding machines or artificial intelligence from being considered inventors. This interpretation aligns with the traditional understanding of invention as a human-centric activity, ensuring that the title of "inventor" is reserved for natural persons, not machines or AI entities.

Inventions by AI

The issue of patentability in the context of inventions identified and solved by AI has been a subject of considerable legal debate. The USPTO has established clear guidelines indicating that for an invention to be patentable, there must be significant human involvement, particularly in the identification of the problem and the creation of the solution.

A landmark decision by the USPTO in April 2020 concluded that an AI, specifically an AI named DABUS, could not be a named patent inventor. This decision was based on the understanding that a patent may only be issued to a natural person. The case of DABUS, which was designed to generate a fractal light signal with pulses to match human brain waves, underscored this principle. The USPTO's reasoning was grounded in patent statute language and Federal Circuit decisions, emphasizing that the "inventor" must be a human [1]

The implications of this ruling are significant. It suggests that even if AI is used in the process of invention, the critical inventive leap - the novel and non-obvious innovation - must be attributed to a human. This human-centric approach to patent inventorship is based on the interpretation of results and decisions made by a human, not by the AI. For instance, if an AI suggests multiple new processes, the human who reviews, selects, and implements the most effective solution based on their training and experience would be considered the inventor.

U.S.C Title 35 in 2024 and Beyond

Aim of U.S.C Title 35 is to promote progress by protecting intellectual property, thereby encouraging innovation and investment in new ideas. A fundamental aspect of this law is its emphasis on human inventorship, a principle rooted in policy and societal reasons, acknowledging the unique human capacity for creativity and problem-solving. The law repeatedly refers to inventors as natural persons, as seen in its language such as “whoever invents or discovers” (35 U.S.C. § 101) and the use of personal pronouns in 35 U.S.C. § 115. This human-centric approach aligns with the broader goal of fostering individual ingenuity and ensuring that the benefits of inventions accrue to human creators.

The question of whether AI can be recognized as an inventor has been a subject of legal debate globally. For instance, the UK Court of Appeal has ruled [2] that AI cannot be the inventor of new patents, emphasizing that only a natural person can hold patent rights. Similar decisions have been made in several other countries, reaffirming the notion that patents are a statutory right granted to persons, not machines. However, the legal landscape is not entirely uniform. In Australia, there has been a ruling that inventions created by AI can qualify for a patent, though this remains an outlier.

Patent Office in 1802

Law academia has argued that the law attributing inventions exclusively to humans is outdated, suggesting an ongoing evolution in the legal understanding of AI and inventorship. This diversity in international legal perspectives underscores the evolving nature of patent law in the age of AI and the ongoing debate about the role of machines in the creative process.

Patent Board from 1790 with Thomas Jefferson, Henry Knox, and Edmund Randolph.

While the AI can assist in the process of invention, the ultimate credit lies with, and goes to the human, who needs to remain in charge and supervise both the identification of the problem and solution for the invention. The amount of supervision may not be important as long as the human is in-charge. This human-centric approach is vital not only for maintaining the integrity of the patent system but also for ensuring that the rewards and responsibilities associated with a patent are borne by humans, who have the capacity for ethical and moral judgment. For example, AlphaFold by DeepMind, which can generate novel and potentially patentable protein structures, underscores this thesis. These generated protein structures can be patentable as long as effort is made on human supervision to craft the problem and later be involved in selection of the protein structure to some extent. The involvement of human supervision in defining the problem and in the selection process of the protein structure is essential to meet the criteria for patentability under the current legal framework, which is fundamentally designed around human creativity and innovation.

References

Further read

LLM, Us and Neuroplasticity

"Are you sure?": Unveiling the Quest for Human-Like Artificial Intelligence

Designers, creativity and AI

A LLM as a reflection of our inner most desires

Should AGI be a she, he, or a it?