In 2017, something revolutionary happened, and yet hardly anyone took notice. In March of this year, the Luxembourg branch of the world’s oldest music collecting society, the Société des auteurs, compositeurs et éditeurs de musique (SACEM), for the first time recognised an AI as an author and registered the musical works it produced for use. This is an unprecedented process, as it was previously legally undisputed that only natural persons, i.e. humans, could produce intellectual creations within the meaning of copyright law. This part of the series therefore examines whether AI can be regarded as an intellectual creator under existing copyright law, and what conditions would have to be met for it to be recognised as such.
AI in the Music Industry – Teil 15: Is AI a Creator?
The artificial intelligence that has been granted this human privilege is called AIVA, which stands for “Artificial Intelligence Virtual Artist”, and was created by Luxembourg-based AI company AIVA Technologies with funding from the EU and the Luxembourg government.[1]
AIVA as a Creator
The brothers Pierre and Vincent Barreau, Denis Shtefan and Arnaud Decker had founded AIVA Technologies in the Grand Duchy of Luxembourg a year earlier with the aim of using AI tools to compose music for films, commercials and games. This is production music, which, as we outlined in an earlier section, is intended to be produced cheaply and in which AI was used at an early stage. The special feature is that the “music machine” AIVA has been trained using a deep learning process with 30,000 works of classical music by the most famous composers to produce short musical works. Initially, these works were registered by AIVA Technologies with SACEM for human composers, until AIVA itself was granted composer status by the collecting society in March 2017. To date, 37 musical works have been registered with the SACEM Luxembourg as authors for AIVA, as a database search (as of February 2023) has revealed, including the ‘Fantasie Symponique ‘Le Reveil’ op 7′, the ‘Celtic Dance op 14 in A minor’ and the overture ‘AIVA – ‘Letz make it happen’ op 23′, which was premiered at the Luxembourg Philharmonie on 23 June 2017 to mark Luxembourg’s National Holiday.[2]
We can only speculate as to why AIVA was granted copyright personality by SACEM and other similar music AIs were not. Perhaps it has something to do with the fact that AIVA was the result of an EU research project in Luxembourg, and the SACEM branch there saw this as an advantage and marketing coup. In any case, AIVA Technologies received EUR 1.5 million in funding from Chinese tech giant NetEase in June 2020[3] and also offers a commercial composition tool that can be used to compose not only film music, but also pop, rock and jazz tracks.[4]
Locke and Kant on Creatorship
Recognising an AI as the author of music raises many questions that shake the foundations of copyright law as a whole. The legitimacy for the introduction of a copyright on intellectual achievements can be traced back to John Locke’s natural law theory of labour in the late 17th century and Immanuel Kant’s theory of human personality rights in the 18th century.[5] In his “Second Treatise of Government” Locke postulates that man is the owner of himself and concludes “The labour of his body, and the work of his hands, we may say, are properly his. Whatsoever then he removes out of the state that nature hath provided, and left it in, he hath mixed his labour with, and joined to it something that is his own, and thereby makes it his property. It being by him removed from the common state nature hath placed it in, it hath by this labour something annexed to it, that excludes the common right of other men: for this labour being the unquestionable property of the labourer, no man but he can have a right to what that is once joined to, at least where there is enough, and as good, left in common for others.”[6] From this it can be concluded that everything man produces through his labour belongs to him. Interestingly, no natural law theory of intellectual property can be directly derived from the labour theory, because Locke had in mind processes of material appropriation through labour, but the argument does not imply any limitation to material property. However, it is crucial that Locke links property to human labour, regardless of whether it produces material or immaterial goods. In this sense, therefore, intellectual property can only arise through human labour.
Another point of reference for justifying a right to intellectual property is Kant’s theory of personality. In his little-known work ‘Von der Unrechtmäßigkeit des Büchernachdrucks’ (‘On the Unlawfulness of Reprinting’) from 1785, the Königsberg philosopher draws a clear line between the material good of a book and the immaterial intellectual achievement that takes the form of a text in a book. Kant argues that the author speaks to the reader through his text and that a part of his personality is reflected in it.[7] The personality of the author is thus communicated to the reader in the text. This has led to the idea that the work should be understood as an ideal part of a natural person.
Thus, Locke’s theory of labour and Kant’s theory of personality became the two main pillars for the justification of authors’ rights under the law of personhood and the law of property. In particular, moral rights are intended to protect the intellectual and ideal link between authors and artists and their works or performances. It follows that copyright can only be granted to natural persons and never to legal persons, animals or machines. For this reason, in the majority of continental European legal systems, copyright is linked to the capacity for personal intellectual creation. Although an AI can independently produce unpredictable results, this is not sufficient to qualify as a work worthy of copyright protection.
The legal situation is different in Anglo-American legal systems, which allow copyright regardless of the intellectual creation, as in the case of work-for-hire contracts. In such cases, the copyright belongs to the client, not the actual author.[8] But even in the Anglo-American legal system, court decisions have repeatedly confirmed that only human beings can be authors of intellectual works.
Naruto as a Creator?
The so-called “monkey selfies” are a particularly curious case.[9] In 2008, British nature photographer David Slater travelled to the rainforests of Indonesia to take a series of photographs of wild Celebes black macaques. Unable to take portraits of the monkeys, he arranged his photographic equipment so that the macaques could take photographs of themselves. In 2011, Slater licensed the photos to the Craters News Agency, which published some of them in British newspapers. When a Wikimedia Commons editor posted the selfie photos published by the ‘Daily Mail’ online in July 2014, he assumed the photos were in the public domain because they were taken by monkeys, not humans. A few days later, Slater discovered the photos on Wikimedia and demanded that the Wikimedia Foundation take them down immediately, citing his copyright. The Caters News Agency backed Slater’s request, but Wikimedia refused to take them down on the grounds that they were in the public domain. The dispute dragged on for three years, and Slater sued for the GBP 10,000 in damages he claimed the publication had caused him. Finally, the US Copyright Office intervened and issued a clarification in December 2014: “To qualify as a work of ‘authorship’ a work must be created by a human being (…). Works that do not satisfy this requirement are not copyrightable. The Office will not register works produced by nature, animals, or plants.”[10] The case, which already seems strange enough, became even more absurd in September 2015 when the People for the Ethical Treatment of Animals (PETA) filed a lawsuit against David Slater for infringing the copyright of the macaque, who was christened Naruto by the animal rights activists. Both the court of first instance and the court of appeal found that animals were not entitled to copyright and acquitted Slater of copyright infringement.[11] However, the curious legal dispute over a monkey selfie shows that in both the US and the UK, only humans can create copyright. Accordingly, an AI cannot be a copyright holder.
Even if an AI is not currently considered to be the author of a musical work, it is worth conducting a thought experiment: what if it were? What are the requirements for an AI to be considered the author of a musical work? The most important characteristic of a work to be copyrighted is the originality of the work. Copyright could be granted to an AI if it could be shown that, firstly, the performance (the output) was produced independently, i.e. without the guidance of a human being; secondly, the performance is peculiar or original; and thirdly, it is a personal intellectual creation, i.e. creative. Applying this strict standard for copyright protection, none of the software AI systems in circulation meet these criteria. Whether it’s David Cope’s Emily Howell AI or AIVA from the Luxembourg company of the same name, they all require human instructions to produce an output. The output may appear to be original, but it is still only the result of pseudo-creativity. In all these cases, the AI is a human tool and, ultimately, the copyright should go to the human who gave the instructions, provided that the output is original.
Machine Consciousness?
Google’s WaveNet, on the other hand, is an unsupervised learning AI system that, after processing the training data, produces musical works and audio recordings on its own, without human guidance. In any case, the result can be considered original if no similarities to existing works can be detected. The question remains whether this is a creative process and whether the AI can develop a personality.
AI researchers agree that this would require awareness processes in AI, which are technically limited in existing software-based AI systems, as AI researcher Ralf Otte explains: “The software AI commonly used today has quite narrow limits, because in the end it’s always just algorithms on a digital computer that run during an analysis” [translation by the author].[12] Nevertheless, Otte can imagine something like machine consciousness. The technical prerequisite is the technology of neuromorphic computers that synthetically replicate the neurons and synapses in the human brain and create artificial neural networks.[13] They differ from current software AI in that they process not only numbers using algorithms, but also electrical impulses, known as spikes, just like the brain. They are therefore known as spiking neural networks (SNNs).[14] These systems operate in analogue mode, i.e. the electrical impulses are processed directly and not converted into ones and zeros as in digital systems. There are also purely digital and analogue-digital hybrid variants of SNNs. However, analogue SNNs in particular have the advantage of being faster, more energy-efficient and better at mapping neural processes in the brain than software AI.[15] Although neuroprocessors are still in the research phase, hundreds of millions of euros are being invested into these new AI systems in Germany alone.[16]
Otte believes that AI systems will soon not only be able to see and hear, but also to perceive internally, i.e. to analyse internal states. This would be a prerequisite for machine consciousness.[17] However, Otte warns against equating this state analysis with feelings. He is convinced that machine consciousness cannot develop feelings or a will of its own, as he explains in detail in his book. However, it can be concluded that AI systems could develop machine consciousness in the foreseeable future, which will raise new legal, moral and ethical questions.
Endnotes
[1] See EU project database CORDIS, “AI that composes complex instrumental music for movies, games, advertising and other types of digital media”, August 17, 2022, accessed: 2024-05-13.
[2] See YouTube, “AIVA – ‘Letz make it happen’ op. 23”, June 29, 2017, accessed: 2024-05-13.
[3] Music Business Worldwide, “Tencent rival NetEase invests $1.7m in Luxembourg-based AI music firm, AIVA”, June 17, 2020, accessed: 2024-05-13.
[4] AIVA, “AIVA – Your personal AI music generation assistant”, n.d., accessed: 2024-05-13.
[5] See chapter 3.2 “Lockes Arbeitstheorie” (pp 76-96) as well as chapters 6.4.3 “Persönliches Recht (Kant) (pp 398-405) and 6.4.4 “Das unveräußerliche Individuelle” (pp 405-415) in Eckhard Höffner, 2011, Geschichte und Wesen des Urheberrechts, vol 1, 2nd edition, München: Verlag Europäische Wirtschaft.
[6] John Locke, 2005, Second Treatise of Government, sect. 27, digitized by Project Gutenberg: https://gutenberg.org/ebooks/7370.
[7] For the German original citation see Immanuel Kant, 1923, “Von der Unrechtmäßigkeit des Büchernachdrucks”, S. 81 in: Kants Werke, vol VIII “Abhandlungen nach 1781”, edited by Heinrich Maier, Berlin: Königlich Preußische Akademie der Wissenschaft.
[8] US Copyright Act, 17 U.S.C. 101 – Definitions.
[9] A detailed description of the case can be found on Wikipedia, “Monkey selfie copyright dispute”, October 23, 2023, accessed: 2024-02-08.
[10] United States Copyright Office, 2014, Compendium of U.S. Copyright Office Practices, December 22, 2014, p 22.
[11] United States Court of Appeals for the Ninth Circuit, Naruto v. Slater, 888 F.3d 418 (9th Cir. 2018).
[12] Ralf Otte, 2021, Maschinenbewusstsein. Die neue Stufe der KI – wie weit wollen wir gehen? Frankfurt/New York: Campus Verlag, Kindle edition, pos 1452.
[13] Ibid., pos 2092.
[14] Ibid.
[15] Ibid., pos 2131.
[16] Ibid., pos 2109-2131.
[17] Ibid., pos 2834-2862.
Music Business Research 
It’s becoming clear that with all the brain and consciousness theories out there, the proof will be in the pudding. By this I mean, can any particular theory be used to create a human adult level conscious machine. My bet is on the late Gerald Edelman’s Extended Theory of Neuronal Group Selection. The lead group in robotics based on this theory is the Neurorobotics Lab at UC at Irvine. Dr. Edelman distinguished between primary consciousness, which came first in evolution, and that humans share with other conscious animals, and higher order consciousness, which came to only humans with the acquisition of language. A machine with only primary consciousness will probably have to come first.
What I find special about the TNGS is the Darwin series of automata created at the Neurosciences Institute by Dr. Edelman and his colleagues in the 1990’s and 2000’s. These machines perform in the real world, not in a restricted simulated world, and display convincing physical behavior indicative of higher psychological functions necessary for consciousness, such as perceptual categorization, memory, and learning. They are based on realistic models of the parts of the biological brain that the theory claims subserve these functions. The extended TNGS allows for the emergence of consciousness based only on further evolutionary development of the brain areas responsible for these functions, in a parsimonious way. No other research I’ve encountered is anywhere near as convincing.
I post because on almost every video and article about the brain and consciousness that I encounter, the attitude seems to be that we still know next to nothing about how the brain and consciousness work; that there’s lots of data but no unifying theory. I believe the extended TNGS is that theory. My motivation is to keep that theory in front of the public. And obviously, I consider it the route to a truly conscious machine, primary and higher-order.
My advice to people who want to create a conscious machine is to seriously ground themselves in the extended TNGS and the Darwin automata first, and proceed from there, by applying to Jeff Krichmar’s lab at UC Irvine, possibly. Dr. Edelman’s roadmap to a conscious machine is at https://arxiv.org/abs/2105.10461