PatentNext Summary: In August 2022, the Federal Circuit in Thaler v. Vidal held that U.S. patent law requires a “human” inventor. In 1884, the U.S. Supreme Court in Burrow-Giles Lithographic Co. v. Sarony held that a human could be the “author” of a photograph. In both cases, a type of “machine” was used to produce an output. Namely, in Thaler a “creativity machine” (a type of Artificial Intelligence (AI)) output and conceived the subject matter of a patent application allegedly without human involvement. In Burrow-Giles a camera output (captured) a photograph, but with human involvement (i.e., the photographer selected and arranged the scene of the photograph). While the Thaler court found that a machine alone cannot be an inventor, it raised the possibility of AI-assisted inventorship that included human involvement (e.g., sole or joint inventorship). Future courts considering the issue of AI inventorship could find the 1884 case of Burrow-Giles instructive, where the photographer’s selection and arrangement for the scene of a photograph in 1884 provides a useful analogy to a modern-day human AI developer’s selection and arrangement of training data, hyperparameters, or other features typically required to train and/or use an AI model.

****

The Federal Circuit’s recent Decision on Artificial Intelligence (AI) Inventorship

In August 2022, the Federal Circuit recently confirmed that U.S. patent law requires that an “inventor” be a natural person (i.e., a “human being”). See  Thaler v. Vidal, Case No. 2021-2347 (Fed. Cir. Aug. 5, 2022). 

Specifically, in Thaler, the Federal Circuit considered whether a “creativity machine” – i.e., an artificial intelligence (AI) machine named “DABUS” (short for “Device for the Autonomous Bootstrapping of Unified Sentience”) – could be an “inventor.” The Thaler court looked to Section 100 of Title 35, which provides “Definitions” for patent-related law.  Because Section 100(f) defines an “inventor” as an “individual,” the Thaler court found that an inventor must be a natural person (i.e., a “human being”) in accordance with current U.S. law, including Supreme Court jurisprudence defining an “individual” as a human. 

The Thaler court also based its decision on Section 100(g), which defines a “joint inventor” (and “coinventor”) as “any 1 of the individuals who invented or discovered the subject matter of a joint invention.” 35 U.S.C. § 100(g) (emphasis added). The Thaler court found that this section further confirmed a human inventor requirement. 

The Thaler court dismissed arguments that other sections of Title 35 (such as Section 103) could give rise to AI inventorship. For example, even though Section 103 states that “[p]atentability shall not be negated by the manner in which the invention was made,” the Thaler court found that Section 103 “relates to how an invention is made and does not trump a provision [i.e., Section 100] that specifically addresses who may be an inventor.” Thaler at 8-9 (original emphasis in italics).

Thus, the Thaler court held that an AI machine (such as DABUS) could not be an “inventor” under current U.S. patent law. This decision was not surprising and mirrors the prevailing position adopted by foreign courts or similar tribunals interpreting AI inventorship under respective patent laws worldwide. See Can an Artificial Intelligence (AI) be an Inventor?

As a curious aside, the Thaler court raised the possibility of AI-assisted inventorship, where, for example, a human using AI could lead to inventorship under U.S. law :

[W]e are not confronted today with the question of whether inventions made by human beings with the assistance of AI are eligible for patent protection.

Thaler at 10. 

Thus, it would be a matter of “first impression” for the Federal Circuit (or any U.S. court for that matter) to consider this issue. 

As chance would have it, a U.S. Supreme Court case from 1884 that involved a matter of first impression for camera technology addressed a similar issue, with analogous facts, and that may one day be cited to help determine inventorship for modern-day AI inventions.

The 1884 Supreme Court Decision regarding Camera Technology and Photographs

It was 1884, and cameras, and the photographs they took, were relatively new to the world. In this context, a matter of “first impression” arose at the Supreme Court in a dispute regarding whether a photograph, as taken by a camera (i.e., a machine), was or was not a protectable “writing” or “production” of an “author” in accordance with the then-prevailing copyright law. See Burrow-Giles Lithographic Co. v. Sarony, 111 U.S. 53, 56 and 60 (1884).

At the time, photographs were not expressly considered as works of authorship – a fact that the Burrow-Giles court expressly acknowledged.  See Burrow-Giles at 58 (“The only reason why photographs were not included in the extended list in the act of 1802 is, probably, that they did not exist, as photography, as an art, was then unknown, and the scientific principle on which it rests, and the chemicals and machinery by which it is operated, have all been discovered long since that statute was enacted.”).

In Burrow-Giles, the photograph-in-question was titled Oscar Wilde no. 18 and depicted Oscar Wilde posing in a scene specifically arranged and selected by the plaintiff – Napoleon Sarony, who was a well-known photographer who had taken the photograph (depicted below).

Mr. Sarony accused defendant Burrow-Giles Lithographic (“Burrow-Giles”) of copyright infringement where Burrow-Giles had used the photograph in allegedly unauthorized lithograph reproductions. 

There was no dispute that a human (Mr. Sarony) had operated a camera (a type of “machinery”) to take the photograph. Burrow-Giles, 111 U.S. at 60. Instead, defendant Burrow-Giles argued in its defense that a photograph could not be copyrighted because it was not authored by a person. That is, Burrow-Giles argued that: “a photograph being a reproduction, on paper, of the exact features of some natural object, or of some person, is not a writing of which the producer is the author.” Burrow-Giles, 111 U.S. at 56 (emphasis added).

Mr. Sarony countered that he was an “author” eligible for copyright protection by virtue of his use of his mental conception to arrange and select the scene for the photograph, including positioning Oscar Wilde in a specific pose in front of the camera; arranging the costume, draperies, and other accessories; and, determining the light and shade for the photograph. Id. at 60.

The Supreme Court agreed with Mr. Sarony. In particular, the Supreme Court found that Mr. Sarony’s activities – including arranging, selecting, and determining the scene for the photograph – gave rise to an original work of authorship pursuant to copyright law, even though the camera (a machine) ultimately took the photograph. As stated by the Supreme Court:

[P]laintiff made the [photograph] entirely from his own original mental conception, to which he gave visible form by posing the said Oscar Wilde in front of the camera, selecting and arranging the costume, draperies, and other various accessories in said photograph, arranging the subject so as to present graceful outlines, arranging and disposing the light and shade, suggesting and evoking the desired expression, and from such disposition, arrangement, or representation, made entirely by plaintiff, he produced the picture in suit.

These findings, we think, show this photograph to be an original work of art, the product of plaintiff’s intellectual invention, of which plaintiff is the author, and of a class of inventions for which the constitution intended that congress should secure to him the exclusive right to use, publish, and sell ….

Burrow-Giles, 111 U.S. at 60.

The Burrow-Giles court also noted that “[w]e entertain no doubt that the constitution is broad enough to cover an act authorizing copyright of photographs, so far as they are representatives of original intellectual conceptions of the author.” Burrow-Giles, 111 U.S. at 58.

Application of Burrow-Guiles to modern-day AI Inventorship Issues

Fast forward to 2022, and AI is the new technology, just as cameras and photography were in 1884. And, just as cameras are a type of “machine” used to generate an output, i.e, a photograph (which were analogized to “paintings” in Burrow-Giles), so too can an AI model be configured to be a type of “machine” used to generate an output. For example, in some cases, this can include paintings or images. See, e.g., OpenAI’s DALL E 2 (a type of “AI system that can create realistic images and art from a description in natural language”).

With respect to AI, in many cases, a human is involved (and typically required) in the “training” process of an AI model (e.g., computing code that may be used to output predictions or classifications). Such human involvement may include, for example, arranging, selecting, or otherwise preprocessing training data for a “supervised learning” approach to train a given AI model. Similarly, an “unsupervised learning” training approach typically requires a human to select a set of data for identification of classification. 

Still further, a “reinforcement learning” training approach typically requires a human to set up or otherwise define an “environment” and “rewards” that influence how an intelligent agent takes actions in the environment in order to maximize the notion of cumulative rewards and, thus, generate an AI model drive by the agent that can perform a desired (and rewarded) action.  

As yet another example, for deep neural networks (NNs), a human is typically involved in the selection of “hyperparameters” used to control the learning process, where these can include aspects such as the model architecture (e.g., the number of hidden layers), learning rate, the number of training epochs, and which activation function(s) to use, among many others. 

Thus, human involvement in the selection and arrangement of training data, selection of hyperparameters, or other preprocessing activities for developing or using an AI model draws a parallel to a photographer’s involvement in the selection and agreement of a scene for a photograph. Said another way, an analogy exists between the selection, arrangement, and determination of a human in designing, configuring, and/or using an AI-related model and the selection, arrangement, and determination of a human’s use of a camera to generate a photograph, as illustrated by the Burrow-Giles Court in 1884.

With respect to a patent, a human may be considered an “inventor” or “joint inventor” by virtue of his or her selection, arrangement, and/or determination of training data, hyperparameters, and/or input into a trained model for generating a given output. Because selecting or arranging training data or input is a typical activity for developing AI models, it could be argued to be the modern-day equivalent of selecting, arranging, shades, and lighting for a scene of photographs taken by a camera. 

The analogy would hold, for example, where a human inventor contributed to the conception of a patent invention by virtue of contributing to one of these aspects (e.g., selection of training data), and where such aspect was recited as a claim element. In such cases, human inventorship could arise by the human contributing his or her “mental conception” to the claimed invention per Burrow-Giles – in sum, contributing to the “idea” of the patent invention as required by law. 

Of course, some may argue that Burrow-Giles analogy does not hold because the Burrow-Giles case involved copyright law, instead of patent law.

However, the Burrow-Giles court drew a comparison between patents and copyrights in its decision. In particular, the Burrow-Giles court noted that the mental creativity of a copyrighted work deserves more scrutiny compared to a patent, given that a copyrighted work is less scrutinized at the time of filing with the library of congress than when a patent is during prosecution at the patent office.  Burrow-Giles at 59-60 (“however, to the kindred subject of patents for invention, they cannot, by law, be issued to the inventor until the novelty, the utility, and the actual discovery or invention by the claimant have been established by proof before the commissioner of patents. … Our copyright system has no such provision for previous examination by a proper tribunal as to the originality of the book, map, or other matter offered for copyright. … It is therefore much more important that when the supposed author sues for a violation of his copyright, the existence of those facts of originality, of intellectual production, of thought, and conception on the part of the author should be proved than in the case of a patent-right. In the case before us we think this has been done.”). 

Thus, the mental conception of a photographer (e.g., the selection and arrangement of the given scene) as demonstrated in Burrow-Giles, was sufficient in that case to convey “inventiveness” and “authorship” even though a camera-generated the photograph at issue. 

The same concept could apply for AI-assisted inventorship, where, presumably, a human using AI could give rise to inventorship under U.S. law, as invited by the Federal Circuit in Thaler. That could be either via identifying the human alone as the sole inventor, which would parallel the “authorship” finding in Burrow-Giles. Alternatively, the human might be identified as a joint inventor, which would step closer to the decision in Thaler (but albeit be different). 

In any event, it is likely only a matter of time before we see a court take this issue up. When it happens, the case of Burrow-Giles should be considered, which could give rise to the curious happen-stance where an 1884 case (that considered 1884 technology) can inform a modern-day case considering new-age technology.  

****

Subscribe to get updates to this post or to receive future posts from PatentNext. Start a discussion or reach out to the author, Ryan Phelan, at rphelan@marshallip.com or 312-474-6607. Connect with or follow Ryan on LinkedIn.