Art, said Stephen, is the human disposition of sensible or intelligible matter for an esthetic end.”
― James Joyce, A PORTRAIT OF THE ARTIST AS A YOUNG MAN, Chapter V
[T]he application…identified the author of the Work as the ‘Creativity Machine,’ and noted it was ‘Created autonomously by machine.’
—Complaint, paragraph 17 in Thaler v. Perlmutter, Civ. Action No. 22-01564
McCoy: Dunsel? Who the blazes is Captain Dunsel? …What does it mean, Jim? (Kirk leaves the bridge) Spock? What does it mean?
—The Ultimate Computer, STAR TREK, THE ORIGINAL SERIES
In my recent attempt at spring cleaning, I mentioned that “the Copyright Office’s ‘refusal to register a two-dimensional artwork claim in the work titled ‘A Recent Entrance to Paradise’ (‘Work’).” I also observed that “[e]ven with Roombas and Creativity Machines doing their jobs well, these issues remain a little messy and will need to be considered further.” I just thought I would have a little longer respite.
But that is not the case because the party denied registration, Dr. Stephen Thaler, filed suit on June 2, 2022, challenging that denial. So, we are right back to confronting the muddled questions around what makes something copyrightable, arising now in the context of a contested case concerning a:
copyright application [that] identified the “author” of the Work as the “Creativity Machine,” and Steven Thaler was listed as the owner of that machine who “was ‘seeking to register this computer-generated work as a work-for-hire to the owner of the Creativity Machine.’” The application was rejected because it “lacks the human authorship necessary to support a copyright claim,” and Thaler appealed that rejection to the Review Board of the United States Copyright Office (“Board”). In a recent, seven-page, single-spaced ruling, the Board affirmed that rejection, finding that “the nexus between the human mind and creative expression” remains “a prerequisite for copyright protection.”
[Flynn, Spring Cleaning: Decluttering From Recent Intellectual Property Detritus & Dusting Off Old Posts]
To address this question, let us first look at what Thaler is contending.
At its base, Thaler argues that he owns the copyright in the Work, but a computer was the author of the Work. Thaler’s complaint admits that he developed the advanced artificial intelligence system that generated the work in question. Complaint, par. 1 [As an aside, the Work at issue is entitled “A Recent Entrance to Paradise,” which sounds a bit like an AI-generated title to replace Stairway to Heaven, which has also appeared in these posts before. Of course, it isn’t clear if Thaler’s AI program generated just the Work, or the title as well]. But, despite initiating the creative process, Thaler pleads that the AI system itself “generat[es] creative output …under conditions in which no natural person contributed to the work,” and that the work in question here was “[c]reated autonomously by machine,” “lacked traditional human authorship,” and that “AI can autonomously create works…” Complaint, par. 1, 14, 17, 20, 36. Thus, as expressed in one reading of this post’s title, Thaler argues that the artist that created the Work in question is the recently developed (read “young”) code from which the work sprang.
But Thaler’s analogy seems off. The AI is not the maker of the Work but in some sense, the medium or tools of the art, no different than oil paints applied by the artist, brushes handled by the painter, or a camera pointed by the photographer. Thaler’s affirmative pleading admits his own agency in the enterprise when he notes that he “owns and operates the AI which created The Work,” that he “built and controlled the AI which generated The Work,” that “The Work was only created by the AI at Dr. Thaler’s insistence,” and “The Work only exists due to Dr. Thaler’s investment.” Complaint, par. 47, 52 (emphasis added). Though not affirmatively pled expressly, one can assume that Dr. Thaler’s “insistence” involved more than verbal encouragement and included the physical acts of pressing the power button and various keystrokes without which there would be no Work created. In other words, like biblical creation itself stemming from a First Cause, the Work “can only be accounted for by the existence of a first cause; this first cause, however, must not be considered simply as the first in a series of continuing causes, but rather as first cause in the sense of being the cause for the whole series of observable causes.” Indeed, despite Thaler’s arguments that the AI created the Work, “the observable order of causation is not self-explanatory,” as one steeped in the philosophy of Aquinas might say. Nor is it self-actualizing. Thaler’s operation, control, insistence, and investment deployed and directed the Creativity Machine.
In other words, there would be no work without Thaler and his contributions. Consequently, though he wishes to maintain from the outset that “no natural person contributed to the work as a traditional author,” Complaint, par. 1, it, he made inventive contributions and started the inventive process. The vast amount of work that goes into the development of such sophisticated AI surely meets the Feist standard for originality necessary for copyrightability—”that originality requires independent creation plus a modicum of creativity.” Feist Publications, Inc. v. Rural Telephone Service Co., 499 US 340, 346 (1991). Even at his most humble, Dr. Thaler would have to admit that his efforts provided a “’slight addition’ sufficient to qualify as originality,” L. Batlin & Son, Inc. v. Snyder, 536 F. 2d 486, 493 (2d Cir. 1976), or that the process of production he designed and initiated “required great skill” even if “no changes were required.” Id. As Feist noted, “To be sure, the requisite level of creativity is extremely low; even a slight amount will suffice. The vast majority of works make the grade quite easily, as they possess some creative spark, no matter how crude, humble or obvious it might be.” 499 US at 345.
Though I would love to take credit for first observing that “AI-generated work” is a misnomer for what is better described as human-initiated, AI-supported work, I cannot. For example, Dan Burk noted, at page 263, in a 2020 law review article that “there is a rich web of human effort and support behind any AI undertaking. When we pull aside the AI curtain, it becomes apparent that the attribution of authorship for AI-enabled creations is largely an exercise in tracing legal causation. Indeed, the concept of original expression, which is required for copyright authorship, implies a causal chain tracing the origin of fixed expression.”
Some thought experiments illustrate that the human creativity sufficient to justify copyright in an AI-generated work exists. First, imagine an artist who loads her brush with paint so much paint that it drips off the brush onto the canvas in a speckled pattern that is largely gravity-induced. Or imagine an artist who lines up cans of paint along the side of a canvas and tips them over so that paint runs across the canvas. No one should doubt that art combining a first cause with the reality of physics still employs enough human intervention to be deemed copyrightable. See Alfred Bell & Co. Ltd. v. Catalda Fine Arts, Inc., 191 F.2d 99, 104–05 (2d Cir. 1951) (“Having hit upon such a variation unintentionally, the ‘author’ may adopt it as his and copyright it.”). Likewise, imagine who asks three children to dip their feet in separate paint roller trays filled with different colored paints and run across an eight-foot by eight-foot canvas. Wouldn’t that artist’s conception of the enterprise make him the author of that work though he never physically applied the paint? Because this artist retains control over the decision as to when the painting is done and directs the children’s activity, he or she remains the author even though this artist may “may delegate or relinquish some degree of control over the process and contours of the work’s fixation,” as Burk notes (at 279); see also Cmty. for Creative Non-Violence v. Reid, 490 U.S. 730, 751–52 (1989)(“we consider the hiring party’s right to control the manner and means by which the product is accomplished”).
Indeed, the delegation may be wide in scope, and the contributions of the subordinates may be “substantial or critical to the final form of the work” without making those subordinates authors or co-authors, according to Burk (at 301). Whether such delegation involved employing other people (like these children) or machines (like cameras, gears, or punch cards), the author for copyright purposes remains the controlling, human first cause because the “work is ‘fixed’ in a tangible medium of expression … by or under the authority of the author.” See 17 U.S.C. §101. As Burk notes, AI is a tool of the creator:
The simplest answer to such questions is, of course, to treat the AI system as one would any other creative tool and assign copyright authorship to whomever designed, programmed, or deployed the machine to generate the resulting work—just as one would assign copyright authorship to whomever employed a paintbrush, saxophone, or word processor to generate a creative work. There is never any question of assigning authorship to a paintbrush, saxophone, or word processor, despite their direct involvement in the act of expressive creation. Copyright is rather clearly an entitlement assigned to human creators.”
[Burk, Thirty-Six Views of Copyright Authorship, by Jackson Pollock, 58 Hous. L. Rev. 263, 265-66 (2020)]
And as Burk continues, AI is never truly autonomous:
Note that as in the Thirty-Fourth View, in no way does the machine learning system behave volitionally, intentionally, or autonomously. For any given AI system, a human designed and wrote the program that constitutes the machine learning algorithm. One or more humans selected the training data for the algorithm. One or more humans determined the statistical parameters for the program, modulating overfitting or underfitting of the data. Numerous human choices were made in generating the resulting output. If there is an author, it is one or more of the humans who are sufficiently causally proximate to the production of the output. In some instances, there may be joint authors. In some instances, none of them may be sufficiently causally proximate to claim authorship, and there will be no author, as in the case of an errant wind or feral hogs. But the author is never the machine.
[Burk, Thirty-Six Views of Copyright Authorship, by Jackson Pollock, 58 Hous. L. Rev. 263, 319-320 (2020)]
Though Thaler claims that he has somehow been denied his entitled copyright ownership, Complaint, par. 49, it is not because the Copyright Office failed to recognize the human creativity inherent in AI and AT-generated works, but because Thaler affirmatively argued the lack of any human involvement in creating the Work. So Thaler could get it registered by admitting the obvious—AI was a tool that Thaler used to create the Work, making him the author.
But Dr. Thaler does not appear to want to make the best argument to register his intellectual property but to make the best arguments to have AI recognized as the creator of such IP. One need only look at a recent Australian case also involving Thaler, where he was trying to establish an AI program as an inventor entitled to recognition under the patent laws. As that Court noted, in paragraph 119, “in filing the application, Dr. Thaler no doubt intended to provoke debate as to the role that artificial intelligence may take within the scheme of the Patents Act and Regulations.” As that Court went on to observe in paragraph 121, “[i]n the present case, it was said to be an agreed fact that DABUS is the inventor of the invention the subject of the application and that Dr. Thaler is not. However, the characterisation of a person as an inventor is a question of law. The question of whether the application that is the subject of this appeal has a human inventor has not been explored in this litigation and remains undecided.”
Thaler is also making AI patent inventorship claims (in his opening brief and later reply) in a US case, Thaler v. Vidal, Case No. 2021-2347, argued on June 6, 2022 (argument starting at 42:45). Essentially, in claiming in the copyright context in the new lawsuit and in the patent context in this recently argued appeal, that the author or inventor need not be a human being, Thaler seeks to render the human participant irrelevant or unnecessary, a dunsel as it were. These are the same sorts of arguments made in Australia.
In the recently argued patent case, the USPTO summarized the question presented, and its answer, in its brief as follows:
can source code be an “inventor” under the Patent Act? Answering that question in the negative, both the USPTO and the District Court correctly focused on the plain language of the relevant Patent Act provisions, as mandated by decades of authority from both the Supreme Court and this Court. Congress amended the Patent Act in 2011 to include both an explicit definition for the term “inventor” that equated an “inventor” with an “individual” and, in a subsequent statutory provision, a reference to that “individual” using personal pronouns. Applying the Supreme Court’s recent holding on the plain statutory meaning of the term “individual,” and recognizing that Thaler could point to nothing – within the text of the Patent Act or otherwise – demonstrating that Congress had intended a different meaning for the term, the District Court correctly concluded that the plain language of the Patent Act required an “inventor” to be a “natural person” (i.e., “human being”). Unsurprisingly, this Court’s precedent has similarly held that an “inventor” seeking patent protection must be a “natural person.”
Under well-settled Supreme Court authority, it is here that any exercise in statutory construction must “end.”
The artist, then, is not the code, but the coder:
This supreme quality is felt by the artist when the esthetic image is first conceived in his imagination. The mind in that mysterious instant[,]…[t]he instant wherein that supreme quality of beauty, the clear radiance of the esthetic image, is apprehended luminously by the mind which has been arrested by its wholeness and fascinated by its harmony is the luminous silent stasis of esthetic pleasure, a spiritual state very like to that cardiac condition … called the enchantment of the heart.
[James Joyce, A Portrait of the Artist as a Young Man, Chapter V]
Would that it were so simple for all to understand. Some commentators seem to accept Thaler’s misapprehension of the facts. Look at what Mirae Heo has said:
The Copyright Office seems convinced that, for now, a work solely created by AI lacks the required human input necessary to constitute a copyrightable work. Perhaps advancements in AI will reach a point in the future where the Copyright Office will recognize the human intelligence in AI as human authorship.
Her two sentences well illustrate the confusion. Her first sentence is based on the strawman fallacy that there can be “a work solely created by AI.” Because AI is itself the human input, its output is imbued with human creativity necessary for copyright registration. That, in turn, seems to be the conclusion in the second sentence, which she leaves in the context of requiring some future recognition by the Copyright Office. But, in reality, the Copyright Office might today “recognize the human intelligence in AI as human authorship.” It is just that Thaler is simply, and purposely, leaving out that argument.
It is unclear why, other than some sense of sport, recognizing AI as itself the inventor or author rather than leaving that recognition to the creator of the AI itself is being pursued so strenuously. That AI might produce a piece of artwork more rapidly than a human painter hardly suggests a reason to raise the AI over its own creator. We have known since 1884 that certain technologies (like photography) generate and capture images much more quickly than any artist painting scenes by hand. See Burrows-Giles Lithographic Co. v. Sarony, 111 U.S. 53, 60–61 (1884). But no one called the box camera that author for copyright purposes.
I have queued in my DVR as sources of answers to life’s mysteries The Godfather and Star Trek, the Original Series. (Not sure why I am revealing this, and I am certainly hoping that I don’t come across like 2d Lt. Jonathon James Kendrick, as he bellowed out about having two books at his bedside, the Marine Corps Code of Conduct and the King James Bible) Leaving the lessons drawn from The Godfather for your later reference, let me turn the Star Trek’s second season episode The Ultimate Computer, an AI fable about the M-5 multitronic unit, a computer capable of “growing, learning” and that was “[t]he most ambitious computer complex ever created. Its purpose is to correlate all computer activity aboard a starship, to provide the ultimate in vessel operation and control.” Though its AI capabilities allowed such growth and learning and even allowed the M-5 to get the Enterprise to have “reacted more rapidly than human control could have maneuvered her,” the M-5 had to be destroyed because it lacked human judgment, compassion, and other intangibles necessary for command. The lesson of that fable was that putting such artificial intelligence in command was “not desirable. Computers make excellent and efficient servants, but I have no wish to serve under them.” That is certainly true in terms of the arts, and the sense of creation they bring. AI can enhance, not replace, supersede, or substitute for, human creation, and it seems a somewhat worthless endeavor to want to call a machine or process a creator.
Even the Star Trek episode in question recognized that the AI-driven M-5 unit was not separate from its creator but an extension of him:
SPOCK: Doctor Daystrom, you impressed human engrams on the M-5 circuits.
KIRK: Whose engrams?
DAYSTROM: Why, mine, of course.
MCCOY: Of course….
[The Ultimate Computer, STAR TREK, THE ORIGINAL SERIES]
It would seem they may be a lesson in that fable.
That is helpful, since, as Pavlyuk, O., et al. note (at page 238), the “difficult battle” that “Plaintiff Thaler was waging” over the creation of art continues. Consequently, answers are still needed. A regime is needed where the invention created through AI may find protection without needing to indulge the fiction that the invention was created by AI. Just as the photographer who points the camera is recognized as the creator of the photo, the programmer who points the program at the problem is a creator of the answering invention yielded through the AI program. While some like Pavlyuk, O., et al., have argued (at page 232) for the express “recognition of the legal personality of artificial intelligence” and others like Russ Pearlman have contended that “United States intellectual property law must recognize AI systems as authors and inventors”(at page 37, emphasis added), I’d rather just gamble on humanity. As Tolstoy said in What Is Art? (at 43), “art is a human activity” and is “indispensable for the life and progress toward well-being of individuals and of humanity.” AI is just another assistant in the studio of master and can help in the creation of new masterpieces(, which in itself still leaves lots of questions for exploration, and such “debate … doesn’t look like it is going to end any time soon”).