Three Laws Blog: Law and Artificial Intelligence

Tu on Use of Artificial Intelligence to Determine Copyright Liability for Musical Works

Shine (Sean) Tu, West Virginia University College of Law, is publishing Use of Artificial Intelligence to Determine Copyright Liability for Musical Works as a WVU College of Law Research Paper. Here is the abstract.

This article explores the use of artificial intelligence to help define the current test for copyright infringement. Currently, the test for copyright infringement requires the jury or a judge to determine whether the parties’ works are “substantially similar” to each other from the vantage point of the “ordinary observer.” This “substantial similarity” test has been criticized at almost every level due to its inconsistent nature. Artificial intelligence has evolved to the point where it can be used as a tool to resolve many of the current issues associated with the “substantial similarity” test. Specifically, courts would no longer have to rely on a battle of the experts or the use of lay observers to determine if a work is substantially similar to another work. Using a new test based on the “ordinary AI observer” copyright infringement can be established using a means that is both less biased and more fact driven while giving alleged infringers a means by which to check ex ante if their work could be infringing.

Download the article from SSRN at the link.

Burk on Thirty-Six Views of Copyright Authorship @UCILaw

Dan L. Burk, UCI School of Law, is publishing Thirty-Six Views of Copyright Authorship, By Jackson Pollock in volume 58 of the Houston Law Review. Here is the abstract.

Humans have long used a variety of tools to convey artistic expression. Perhaps the most recent and mysterious artistic tools are machine learning or “artificially intelligent” (AI) computer systems that have captured popular attention. When taken in isolation, these devices seem to operate autonomously, giving the illusion that there is no author behind their output. In fact, 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. In this article I show that concepts of causation, volition, and intention that are familiar from other areas of law also inform copyright authorship, and that the machine learning revolution affords us the opportunity to reveal previously hidden assumptions about copyright authorship. While I will begin by illustrating these concepts with examples from the graphic arts, the same principles are readily applied to other authorial works in other media. The result dispels not only the confusion surrounding mechanical creation, but a variety of long-standing problems in copyright authorship.

Download the article from SSRN at the link.

Levendowski on How Copyright Law Can Fix AI's Implicit Bias Problem @levendowski

Amanda Levendowski, New York University School of Law, is publishing How Copyright Law Can Fix Artificial Intelligence's Implicit Bias Problem in the Washington Law Review. Here is the abstract.

As the use of artificial intelligence (AI) continues to spread, we have seen an increase in examples of AI systems reflecting or exacerbating societal bias, from racist facial recognition to sexist natural language processing. These biases threaten to overshadow AI’s technological gains and potential benefits. While legal and computer science scholars have analyzed many sources of bias, including the unexamined assumptions of its often-homogenous creators, flawed algorithms, and incomplete datasets, the role of the law itself has been largely ignored. Yet just as code and culture play significant roles in how AI agents learn about and act in the world, so too do the laws that govern them. This Article is the first to examine perhaps the most powerful law impacting AI bias: copyright. Artificial intelligence often learns to “think” by reading, viewing, and listening to copies of human works. This Article first explores the problem of bias through the lens of copyright doctrine, looking at how the law’s exclusion of access to certain copyrighted source materials may create or promote biased AI systems. Copyright law limits bias mitigation techniques, such as testing AI through reverse engineering, algorithmic accountability processes, and competing to convert customers. The rules of copyright law also privilege access to certain works over others, encouraging AI creators to use easily available, legally low-risk sources of data for teaching AI, even when those data are demonstrably biased. Second, it examines how a different part of copyright law — the fair use doctrine — has traditionally been used to address similar concerns in other technological fields, and asks whether it is equally capable of addressing them in the field of AI bias. The Article ultimately concludes that it is, in large part because the normative values embedded within traditional fair use ultimately align with the goals of mitigating AI bias and, quite literally, creating fairer AI systems.

Download the article from SSRN at the link.

Guadamuz on a Comparative Analysis of Originality in Artificial Intelligence Generated Works @technollama

Andrés Guadamuz, University of Sussex, is publishing Do Androids Dream of Electric Copyright? Comparative Analysis of Originality in Artificial Intelligence Generated Works in Intellectual Property Quarterly (2017). Here is the abstract.

The advent of sophisticated artificial neural networks has opened new artistic opportunities, but also a variety of new legal challenges. Computer programs such as Google's Deep Dream can take an image and process it in manners that resemble biological networks, producing artwork that is both unique and unpredictable. The law is not unfamiliar with the challenges of artificial intelligence, in the past academics and policymakers have had to deal with the legal implications of autonomous agents in contract formation, just to name one are of interest. However, for the most part the implementation of smart systems has been limited in their reach and scope, and in many instances autonomous agents required quite a lot of direction from the programmer, following a very stringent set of rules. This meant that for the most part all rights, responsibilities and liabilities arising from artificial agents fell squarely on the program creator. Neural networks are different, these systems have the potential to generate works in which human interaction is minimal. Modern copyright law has been drafted to consider originality as an embodiment of the author’s personality, and originality is one of the main requirements for the subsistence of copyright. So, what happens when you remove personality from the equation? Are machine-created works devoid of copyright? Do we need to change copyright law to accommodate autonomous artists? This session will explore this and other questions.

Download the article from SSRN at the link.

Ramalho on a Proposed Model for the Legal Status of AI's Creations @Copytwilight @MaastrichtU

Ana Ramalho, Maastricht University, has published Will Robots Rule the (Artistic) World? A Proposed Model for the Legal Status of Creations by Artificial Intelligence Systems. Here is the abstract.

While currently artificial intelligence (AI) is not completely independent from human input, the speed and direction of technology development seem to anticipate a not-so-distant future where it will be. From a copyright perspective, this scenario challenges traditional conceptions, notably that of authorship. In many jurisdictions, authorship seems to be somewhat connected to the conditions for protection, which might imply that, absent a human author, a work will not be original and therefore not copyrightable. This may leave many works that would otherwise be copyrightable without protection, thereby causing legal uncertainty; but it also raises questions about whether protection should at all be available, and about whether copyright is fit for purpose in face of technological progress in the area of AIS. The first part of this paper focuses on whether the current copyright framework can accommodate AIS as creators for purposes of copyright protection. To that end, I examine and compare requirements of authorship in Europe, the US and Australia. I then analyse current legal constructions that could accommodate non-human authorship, such as the specific regime of computer-generated works existent in some common law jurisdictions. The second part of the paper enquires whether copyright should protect AIs as creators. To answer that question, I look into the different rationales of copyright protection. Building on the data from the previous parts, I propose a model for the legal regime of AIs’ creations.

Download the article from SSRN at the link.

Should AI Have Intellectual Property Rights? @technollama @SussexLaw @ConversationUK

Andres Guadamuz, a professor at the University of Sussex, examines whether AI should have copyright in artwork. He gives as an example a computer that analyzed the work of Rembrandt, and then produced artwork "after" the famous artist. More at The ConversationUK. 

Yanisky-Ravid and Velez-Hernandez on the Copyrightability of Artworks Produced by Creative Robots, Driven by Artificial Intelligence Systems, and the Concept of Originability: The Formality-Objective Model

Shlomit Yanisky-Ravid, Yale Law School; ONO Academic College; Yale University Information Society Project; Fordham School of Law; and Luis Anotnio Velez-Hernandez, Fordham Law School, have published Copyrightability of Artworks Produced by Creative Robots, Driven by Artificial Intelligence Systems and the Concept of Originality: The Formality - Objective Model. Here is the abstract.

In our new era of advanced technology, creative robots, driven by sophisticated artificial intelligence systems and hence, acting autonomously, are capable of producing innovative artistic and other works — ones which, had they been created by humans, would be eligible for copyright protection. This article addresses the question of copyrightability of artworks created by creative robots. We argue that creative robots as autonomous entities are capable of holding copyrights in art works they produce. However, the greatest hurdle to this notion, from the copyright law point of view, is the interpretation of the most important concept in copyright protection: the requirement of originality. Therefore, in this article we are revisiting the concept of originality. This article argues that confronting the challenges of the new 3A era (advanced, automated and autonomy), calls for a reassessment of the meaning of originality, which is undefined in the field of copyright protection both in international law, under the Berne Convention, and in domestic U.S. copyright law. The lack of a clear definition means that the existing concept of originality is inadequate for addressing new, possibly “copyrightable” works produced by creative robots. Moreover, the lack of a clear definition leads to interpretations of “originality” that are vague, immeasurable, and disharmonized, does already cause confusion in the industry as well as the public. This uncertainty surrounding an important legal concept has triggered a search for a solution that could eliminate or, at the very least, reduce future conflicts. This article suggests that a more formal, objective approach — as opposed to the existing, subjective (or mixed) approach — to the concept of originality should be adopted. The proposed objective approach might be applicable to works created by creative robots as well as artworks generated by digital tools and is further warranted by the intangible, vague nature of art. We suggest that a consistent legal determination of the question “what is original?” can successfully be achieved only by a formal, objective, descriptive approach.

Download the article from SSRN at the link.

Yanisky-Ravid and Liu on When AI Systems Produce Inventions: The 3A Era and an Alternative Model for Patent Law

Shlomit Yanisky-Ravid, Yale Law School; ONO Academic College; Yale University, Information Society Project; Fordham University School of Law, and Xianquiong (Jackie) Liu, Fordham University School of Law, have published When Artificial Intelligence Systems Produce Inventions: The 3A Era and an Alternative Model for Patent Law. Here is the abstract.

Currently, robots, Artificial Intelligence and machine learning systems (hereinafter referred to collectively as “AI” or “AI systems”) can create inventions, which, had they been created by humans, would be eligible for patent protection. This study addresses the patentability of these inventions created by AI systems. We argue that traditional patent law has become outdated, inapplicable and irrelevant with respect to inventions created by AI systems. We call on policy makers to rethink current patent law governing AI systems and replace it with tools more applicable to the new (3A) era of advanced automated and autonomous AI systems. Our argument is based on three pillars: the features of AI systems, the Multiplayer Model and the irrelevance of theoretical justifications concerning intellectual property. In order to fully convey the ability of AI systems to create inventions, the article explains, for one the first times in the legal literature, what AI systems are, how they work and what makes them (so) intelligent. This understanding is crucial to any further discourse about AI systems. We identify eight crucial features of AI systems they are: (1) creative; (2) unpredictable; (3) independent and autonomous; (4) rational; (5) evolving; (6) capable of data collection and communication; (7) efficient and accurate; and they (8) freely choose among alternative options. We argue that, due to these features, AI systems are capable of independently developing inventions which, had they been created by humans, would be patentable (and able to registered as patents). The traditional approach to patent law in which policy makers seek to identify the human inventor behind the patent is, therefore, no longer relevant. We are facing a new era of machines “acting” independently, with no human being behind the inventive act itself. The second pillar of our argument is the Multiplayer Model, which characterizes the long process through which inventions are created by AI systems. The Multiplayer Model, which is also almost absent in the current legal publications, describes the multiple participants and stakeholders, both overlapping and independent, involved in the process, including software programmers, data and feedback suppliers, trainers, system owners and operators, employers, the public and the government. The model conveys that the efforts of traditional patent law to identify a single inventor of these products and processes are no longer applicable. The third pillar of our argument is the irrelevancy of theoretical justifications such as personality and inventiveness/efficiency to inventions created by AI systems. In contrast to other scholars, we argue that traditional patent law is irrelevant and inapplicable to these situations, that these inventions should not be patentable at all and that other tools can achieve the same ends while promoting innovation and public disclosure. These other, non-patent incentives include commercial tools such as electronic and cyber controls over inventions, first-mover market advantages and license agreements. This proposal serves a gatekeeping function and is superior to a revision of the non-obviousness standard used by other scholars to afford patent protection to inventions by AI systems. In maintaining the traditional patents system by hunting for a “real” human inventor, policy makers exhibit a misunderstanding of advanced technology and AI systems features. We conclude with a discussion of the implications of our analysis for different legal regimes, such as tort, contracts and even criminal law.

Queensland University of Technology To Hold Regulating Robots Symposium October 7, 2016

The Queensland University of Technology is hosting a symposium on regulating robots, called appropriately enough, the Regulating Robots Symposium. It will be held EVENT: Regulating Robots Symposium October 7, 2016. Here's a description of the event from the QUT website.

As a disruptive technology, robotics is transforming our society and our economy. Robots have been increasingly deployed in innovation as part of Australia’s “Ideas Boom”.

The Intellectual Property and Innovation Law Research Program is holding a Regulating Robots Symposium on Friday, 7 October 2016 from 8:30am to 5pm. To register please email [email protected].

There has been a mixture of hope and anxiety as to how robotics and artificial intelligence will affect jobs, education and employment. This is certainly apparent in the legal profession, In terms of transportation, there are autonomous vehicles, drones, and aquabots.

Robots have been deployed in agriculture, hospitals, and the environment. Robots increasingly feature in civilian law enforcement, and the military battlefield. This symposium focuses upon the regulation of robots, across a range of contexts. The event considers the legal, ethical, and public policy challenges posed by robotics. This event will feature an array of experts from a number of disciplines – including law, philosophy, ethics, public policy, and robotics.

Abbott on AI and the Future of Patent Law

Ryan Abbott, Southwestern Law School and UCLA School of Medicine, has published I Think, Therefore I Invent: Creative Computers and the Future of Patent Law. Here is the abstract.

An innovation revolution is on the horizon. Artificial intelligence has been generating inventive output for decades, and now the continued and exponential growth in computing power is poised to take creative machines from novelties to major drivers of economic growth. A creative singularity is foreseeable in which computers overtake human inventors as the primary source of new discoveries. In some cases, a computer’s output constitutes patentable subject matter, and the computer rather than a person meets the requirements for inventorship. Despite this, and despite the fact that the Patent Office has already granted patents for inventions by computers, the issue of computer inventorship has never been explicitly considered by the courts, Congress, or the Patent Office. Yet the issue of whether a computer can be an inventor is an eminently practical one — not only do inventors have ownership rights in a patent, but failure to list an inventor can result in a patent being held invalid or unenforceable. Drawing on dynamic principles of statutory interpretation and taking analogies from the copyright context, this article argues that creative computers should be considered inventors under the Patent and Copyright Clause of the Constitution. Treating nonhumans as inventors would incentivize the creation of intellectual property by encouraging the development of creative computers. The article proceeds to address a host of challenges that would result from computer inventorship, ranging from ownership of computer-based inventions, to displacement of human inventors, to the need for consumer protection policies. This analysis applies more broadly to nonhuman creators of intellectual property, and explains why the Copyright Office came to the wrong conclusion with its new Human Authorship Requirement. Just as permitting computer inventorship will further promote the progress of science, so too will permitting animal authorship promote the useful arts. Finally, computer inventorship provides insight into other areas of patent law. For instance, computers could replace the hypothetical skilled person that courts use to judge inventiveness. This would provide justification for raising the bar to patentability and address one of the most serious criticisms of the patent system — that too many patents of questionable value are issued. Creative computers may require a whole rethinking of the baseline standard for inventiveness, and potentially of the entire patent system.

Download the article from SSRN at the link.