Shounak Dharap - Law Student, Internet and Intellectual Property Justice Clinic, USF School of Law on
The machines are learning. But unlike HAL 9000’s cold and unapologetic killing of its fellow astronauts, Skynet’s initiation of a nuclear holocaust, or Agent Smith’s tireless quest to eradicate humanity, our machines are more likely to belt out a verse of poetry, play a delightful sonata, or create a captivating watercolor painting. Scientists, researchers, and students have already successfully programmed a wide variety of artificial intelligences that can create hundreds of original compositions of music, stanzas of poetry, and chapters of a novel in a single day. One machine has had hundreds of thousands of its books sold on Amazon. Another machine, Emmy, created five thousand original compositions in the style of Bach in the time it took its creator to go out for a sandwich. Yet another machine, AARON, has created paintings that have been displayed in museums across the world, including at the Brooklyn Museum and the San Francisco Museum of Modern Art. A number of these programs allow a user to enter some input—perhaps a theme—around which the ultimate artistic work is created. The users of these machines are then authors, painters, musicians—artists by proxy, but in their own right; yet they are deprived of the very protections afforded to artists under the law.
Article I, Section 8 of the US Constitution provides that “Congress shall have the power… To promote the Progress of Science and useful Arts, by securing for limited Times to Authors and Inventors the exclusive Right to their respective Writings and Discoveries." Copyright law in the United States is further defined by the Copyright Act, which protects works of authorship that meet the requirements for originality and fixation: works must be independently created by the author and possess some minimal level of creativity; and works must be fixed in a tangible medium of expression. When it comes to originality, the Supreme Court defined original works in Higgins v. Keuffel, 140 U.S. 428, 431 (1991) as those “founded in the creative powers of the mind.”
Under this definition, the program itself is copyrightable as an original product of the programmer’s mind, but its artistic creations are simply products of the program, and not of the user’s mind. The user simply enters in one word or sentence upon which the program builds an original composition. Fixation, too, is a problem for machine-created art. Improvised speeches and non-scripted live events are considered transitory rather than fixed; for a computer program, this means that the artistic output must be consistent and unchanging through multiple iterations. For a program that writes books on its own, fixation may not be such a problem, but a program that creates a new original work based on user input each time it runs will not meet the fixation requirement.
The more compelling question is whether users should even be able to copyright artistic works created by computer programs. The Constitution awards an artist a short monopoly on the artist’s work of art under the assumption that exclusive rights to the work for a period of time after its creation will encourage the creation and publication of those works. However, in the case of a machine that can pump out several thousand original musical compositions over lunch, this incentive is nullified. A machine does not need to be compensated or incentivized to create, and the sheer volume of its creations has the potential to easily overwhelm the sum of all human artistic creation.
Ultimately, allowing users to copyright artistic works created by computer programs based on simple ideas contravenes both the spirit and letter of the Constitution’s Copyright Clause. These machines don’t need an incentive to create. Moreover, denying copyrights to machine-created works will make these artistic works immediately available to society as part of the public domain, a great public benefit.