[Ed Note : In an interesting read, Arvind Pennathur talks about machines which can make artwork, paintings and the like! Is art becoming a science and science becoming an art? Is it possible? Read to find out more.]
Integration of advanced technology into society has become increasingly normal in the past five years as innovative minds constantly push the boundaries of what is achievable and what can be realistically implemented. A large part of this resurgence of complex applied sciences is due to the proliferation of artificial intelligence (abbreviated as AI), into various aspects of everyday life during the 21st century. Ever since its introduction, it has became a staple of the modern world and is being used in a wide variety of ways, right from performing day to day human tasks to performing functions too precise or nuanced for humans, including more creative purposes such as self driving cars[1], news anchoring[2], writing movie scripts[3] and making music that is virtually indistinguishable from the music that is composed by a human[4]. In lieu of these advancements and the pace at which they have come, it is no exaggeration to say that AI is here to stay for the foreseeable future.
One of the areas that has received considerable attention in the recent past is the creation of art pieces by artificial intelligence programs. Though there was initially some apprehension that any serious progress could be made in the field, there has been remarkable growth in technology in the past 4 years. As late as October of this year, Christie’s New York location sold a painting (for $432,500) that was created by an algorithm (named ‘Obvious’) developed by three artists. The program was made to observe thousands of portraits as ‘sample pieces’ in order for it to get an idea on how to go about creating a similar piece[5]. Using characteristics picked up from these images, the program will generate a brad new image[6]. This marked the first time that an art piece made by an AI was seen as traditional art; that is, something to be sold and admired, just as any ordinary piece of artwork. One of the artists behind the AI stated that the target audience was the traditional art market as opposed to those involved in technology[7]. This marked a new, bold direction for artificial intelligence to venture into, with the certainty that such AI-centric methods can be used to produce new works that might contribute to the field of artwork and not to simply demonstrate new applications of technology.
While this may seem like a step too far into the future, it might surprise you to learn that application of artificial intelligence to art is nothing new, and has been around since at least 2014, when the concept of GAN’s (Generative Adversarial Networks) were introduced. GANs’ were modeled after the human brain and designed to produce completely original images, different from the samples fed to them. ‘Obvious’ works on this logic, having been trained to produce pieces that are notably distinguishable from other organically created paintings fed to it, while at the same time retaining their style [8]. The nature of the program used to create the portrait is strikingly similar to the one used to create paintings in the style of the late Dutch painter, Rembrandt Harmenszoon van Rijin. The program ‘learnt’ the style of the painter and subsequently created an independent work that was completely its own but in the style of Rembrandt[9].
This is somewhat similar to another example of the infiltration of AI into the realm of art: Google DeepDream. Introduced in 2015, DeepDream, an image recognition software, uses artificial neural networks that have the ability to emulate how the brain receives and processes information to produce new images by repeating the patterns it perceives in already existing images[10]. However, the factor that separates DeepDream’s neural network from the system that exists in ‘Obvious’ in terms of application is that the latter produces entirely new results that have nothing to do with the inputs that it receives; instead, they are simply used as reference points. Google DeepDream produces an amalgamation of the repetition it observes in the original pieces fed to it.
The challenge is that it becomes difficult to decide on questions of ownership when AI generates original and creative works independent of the humans who created the AI system in the first place[11]. Who is entitled to the licensing rights to the product, to the royalties, and who bears responsibility for copyright infringement and protecting rights from future infringements? Take the Rembrandt project, for example. The amount of work that was put into it is simply staggering, with roughly around 350 paintings being analyzed, and over 150 gigabytes of digitally rendered graphics collected and used in order to provide proper instructions so that the AI could effectively recreate the painting in Rembrandt’s style[12]. It is no wonder that the question of ownership comes up, simply due to the sheer number of people who worked on the project; who would bear responsibility and accountability for the pieces of artwork generated by the system, and what legal rights could he or she assert?
Despite the increasingly rapid advancement in technology, however, there is no law that recognizes the ownership of artificial intelligence. In the United Kingdom, the work of an author has to be ‘original’ in order for it to be protected by the Copyright, Designs and Patent Act[13], as was held in Infopaq International A/S v Dansuke Dagblades Forening[14]. Section 9(3) of the same Act states that the author of anything that is computer generated will be the person ‘by whom the arrangements necessary for the creation of the work are undertaken’[15]. United States law takes a similar stance; in 1979 the Commission on New Technological Uses of Copyrighted Works (“CONTU”) stated that any work that was created with the use of a computer should be afforded copyright protection if it is an original work that falls within the purview of the original 1976 Act[16], which states that the author is the one who conceives of the work and fixes it in a tangible medium of expression. Indian law has the Copyright Act of 1957 that gives the author rights over his creation for his lifetime plus another sixty years[17]. The indication of a definite lifetime makes it obvious that there is no recognition of AI for ownership of artwork. This begs the question: should AI be given legal ownership of their creations?
In order to explore the possible answers to this question, a key area that requires deeper exploration is the capability of an AI to create pure original pieces of work. The programs that have worked thus far such as Google’s DeepDream and ‘Obvious’, all rely on some preliminary input in order for the program to generate competent results. This means that regardless of the lack of human hand in the actual creation of the brand new art piece, it was necessary for the AI to base its creations on already existing works. Essentially, the AI that has been used to create these art pieces at the moment simply ‘functions’ – it does not ‘think’. It does not have any knowledge of what it is being used for, let alone why it has been built; it only serves to create the art piece. This remains the biggest roadblock to recognizing AIs’ as capable owners of their works in the eyes of the law, as they cannot own intellectual property without taking cognizance of the creative process they are employing.
If there comes a situation where an artificial intelligence can form an intention derived from an intrinsic desire and/or belief, and can subsequently act based upon the same, there seems to be a clear cut case of viewing such a program as a legal person[18], statutory definitions aside. However, at the moment, the artificial intelligences that are built to create inventions that can be classified as intellectual property do not autonomously build them – there is always some element of human programming that causes the invention to be created. The intelligence itself does not have the intent of building the invention. In his article “Artificial Consciousness: Utopia or Real Possibility” Giorgio Buttazzo says that despite current technology’s ability to simulate autonomy, it is not possible for computers on their own to exhibit creativity, emotions, or free will[19]. Consciousness involves several facets, one of which is intent. In order to say that an artificial intelligence has a conscience, it has to be proven that its actions are the result of cognitive thinking, which necessitates premonition and cognizance of possible consequences for certain actions.
It would naturally follow that only the most superintelligent AI that has such a consciousness and is able to make artistic pieces on its own would be able to be granted ownership of its creations and will come up with original and creative works[20]. As of today, super intelligent AI has not been developed. However, when such technology is readily available and used in the public domain, perhaps the appropriate laws could be altered so as to allow for them to be creative owners of their works. This would definitely create an impact on the creative processes of any future work, as questions surrounding efficiency and reliability will almost certainly be raised. Can a program create a better painting than a human? Could AI entities purchase artistic pieces of their own? Laws related to sentient AI would have to be carefully constructed as the line between mere ownership of creative work and these entities becoming a permanent force in the market can be easily blurred, and it has the potential to lead to chaotic results.
[1] Andrew J. Hawkins, Uber seeks permission to resume self-driving car testing on public roads, The Verge, https://www.theverge.com/2018/11/2/18056622/uber-self-driving-car-safety-report-testing-pennsylvania
[2] Lucy Handley, The World’s First A.I. News Anchor has gone live in China, CNBC, https://www.cnbc.com/2018/11/09/the-worlds-first-ai-news-anchor-has-gone-live-in-china.html
[3] Annalee Newitz, Movie written by algorithm turns out to be hilarious and intense, ArsTechnica, https://arstechnica.com/gaming/2016/06/an-ai-wrote-this-movie-and-its-strangely-moving/
[4] Brad Merill, It’s Happening: Robots May Be Creative Artists of the Future, MakeUseOf, https://www.makeuseof.com/tag/happening-robots-may-creative-artists-future/
[5] Ahmed Elgammal, With AI Art, Process is more Important than Product, The Smithsonian, https://www.smithsonianmag.com/innovation/with-ai-art-process-is-more-important-than-product-180970559/
[6] Hugos Caselles-Dupré, Obvious, explained, Medium, https://medium.com/@hello.obvious/ai-the-rise-of-a-new-art-movement-f6efe0a51f2e
[7] Naomi Rea, Is the Art Market Ready to Embrace Work Made by Artificial Intelligence? Christie’s Will Test the Waters This Fall, Artnet News, https://news.artnet.com/market/artificial-intelligence-christies-1335170
[8] Tim Schneider and Naomi Rea, Has Artificial Intelligence Given Us the Next Great Art Movement? Experts Say Slow Down, the ‘Field Is in Its Infancy’, Artnet News, https://news.artnet.com/art-world/ai-art-comes-to-market-is-it-worth-the-hype-1352011
[9] Shlomit Yanisky Ravid & Luis Antonio Velez-Hernandez, Copyrightability of Artworks Produced By Creative Robots and the Originality Requirement: The Formality – Objective Model, 19 MINN. J.L. SCI. & TECH.
[10] James Temperton, Create your own DeepDream nightmares in seconds, Wired, https://www.wired.co.uk/article/google-deepdream-dreamscope
[11] Yanisky-Ravid, Shlomit and Moorhead, Samuel, Generating Rembrandt: Artificial Intelligence, Accountability and Copyright – The Human-Like Workers Are Already Here – A New Model, Michigan State Law Review 2017.
[12] Yanisky-Ravid, Shlomit and Moorhead, Samuel, Generating Rembrandt: Artificial Intelligence, Accountability and Copyright – The Human-Like Workers Are Already Here – A New Model, Michigan State Law Review 2017.
[13] Can Copyright subsist in an AI generated work?, Clifford Chance: Talking Tech, https://talkingtech.cliffordchance.com/en/ip/copyright/ai-and-ip–copyright-in-ai-generated-works–uk-law-.html
[14] [2009] ECDR 16.
[15] Copyright, Designs and Patent Act 1988, Section 9(3).
[16] Ryan Abbott, “I Think, Therefore I Invent: Creative Computers and the Future of Patent Law”, B.C.L. Rev. 57(4), 1079 (28 September 2016).
[17] Arya Matthew, Protection of Intellectual Property Rights under Indian and International Law, Altacit Global, https://www.altacit.com/publication/protection-of-intellectual-property-rights-under-the-indian-and-international-laws/
[18] David Calverley, Imagining a Non-Biological Machine as a Legal Person, 22 AI & Society (2008).
[19] Giorgio Buttazzo, Artificial Consciousness: Utopia or Real Possibility, IEEE, July 2001, available at https://pdfs.semanticscholar.org/c505/98f38ae1d10546513166f564e115b06df83e.pdf
[20] Ben Allgrove, “Legal Personality for Artificial Intellects: Pragmatic Solution or Science Fiction?” (June 2004) (Master of Philosophy thesis, University of Oxford), available at https://ssrn.com/abstract=926015