In the past decade, Artificial Intelligence (A.I.) has been used to create artistic works as well as news articles. For example, A.I. has created works which can imitate famous artists such as Rembrandt van Rijn, as well as generating articles for newspapers like The Washington Post. These recent innovations have led some people to ask whether A.I. should be eligible to receive a copyright for its creations. Current Copyright Law does not classify A.I. works as copyrightable creations, however, A.I. might one day achieve a level of intelligence to warrant such accreditation to be protected under United States Copyright Laws.
Since 1973, it has been the official policy of the United States Copyright Office to deny any copyright claims if the work was not created by a human being. In fact, The Compendium of U.S. Copyright Office Practices states that in order for a work to be eligible for copyright protection, it must be “created by a human being.” In one instance, the 9th Circuit Court of Appeals seems supported the requirement that a human must be the creator for a work to receive copyright protection. In the case Naruto v. Slader, the court denied the monkey, Naruto, standing to bring an action under the Copyright Act. This holding from one of the primary goalsof Intellectual Property Law: to benefit society by incentivizing innovation by allowing creators and innovators the right to profit from their creations for a set period.
In his book Superintelligence: Paths, Dangers, Strategies, author Nick Bostrom says that almost all the current A.I. systems used in our society have a “narrow range of cognitive ability.” In other words, what most people consider to be A.I. today are data algorithms designed to analyze a given type of data and produce a certain result. The A.I. used to imitate Rembrandt was designed to perform that sole function, and only did so after analyzing 346 of his paintings. Likewise, IBM’s Chef Watson has created original food recipes, but it was a task the A.I. was created to do and requires 9,000 different recipes to cross reference. Thus, while these A.I. creations are undeniably impressive, they demonstrate that modern A.I. is still a catalyst to creation for human beings rather than creators. The A.I.’s are only as complex or accurate as humanity makes them, and they are incapable of independent creation.
Society has unquestionably benefited from the creative products of A.I, but A.I. innovation would not be affected by any incentives to create copyright eligible material. Each A.I. can be compared to a hammer; it is a tool designed to perform a specific task. The only difference between a hammer and an A.I. is the complexity of the task as neither contains the independent will to choose whether or what to create without which any incentive system would be wasted. Thus, while A.I.s are capable of remarkable creations, they should not be granted copyrights for any work they create until they become capable of independent thought and action.
* Daniel Norton is a second-year law student at Wake Forest University School of Law. He has worked in Information Technology for the past 7 years and is currently working to increase the accuracy of contract analysis Artificial Intelligence. Upon graduation, he intends to practice Intellectual Property Law.