Last month the artwork entitled Portrait of Edmond Belamysold for $432,500 at Christie’s New York auction house. More remarkable than the fact the piece fetched a price more than 40 times its $10,000 estimate, however, is its origin. The portrait is not of human making; it was created by an algorithm defined by the algebraic formula which appears on the work as the artist’s signature. It is the world’s first artwork created by artificial intelligence to be sold at auction.
This coup followed only a fortnight after the, now infamous, sale of Banksy’s Girl with Balloon at auction at Sotheby’s in London when a shredder, concealed by the artist in the picture frame activated as the hammer fell, causing the former picture to ‘self-destruct’ and leading to the creation of the new work Love is in the Bin.
Each event prompted debate as to the artistic merits of the lot sold: Belamy, described in the New York times as evocative of a “17th Century oil portrait, smeared with a damp sponge” divided critics and fans of AI art while the Banksy stunt prompted speculation as to whether this metamorphosis would, in fact, increase the value of the piece compared to its £1m hammer price.
Meanwhile, in the legal world, the lots raise interesting questions as to authorship and ownership of intellectual property rights in the works and the impact of the answers are by no means confined to idiosyncrasies of the art world.
Are the pieces copyrighted?
Any literary, dramatic, musical or artistic work must be original in order to qualify for copyright protection. Notably this is a low bar: artistic merit is not usually required; however, the work must reflect the author’s own intellectual creation. For Banksy’s canvas, remodelled by the author’s design and aided by technology, there are therefore no legal obstacles in considering the new piece as a new copyright work. Sotheby's was pleased have managed the sale of “the first artwork in history to have been created live during an auction".
Belamy too was a new, original creation, generated by an algorithm that learned to imitate 15,000 portraits. Hugo Caselles-Dupré of Obvious, the French art collective behind the algorithm, reflected that portraiture illustrates the collective’s point “that algorithms are able to emulate creativity”.
Who is the author?
The individual or collective who authors a work will normally exclusively own the work and be its ‘first owner’. Importantly, however, the requirement for originality is widely assumed to call for some reflection of personality and therefore a human author. As such it is possible that, despite its originality, any AI work generated with minimal or no human intervention cannot be copyright for want of a human author. Faced with this conundrum, the possible legal solutions are to either deny copyright protection to any work produced autonomously by AI or attribute authorship to the creator of the programme.
Which direction the law takes will be significant not only to tech-savvy artists and creatives but also to businesses wishing to exploit AI commercially for content generation. For the likes of Google, who are already pursuing AI initiatives to create commercial works including a journalistic program to write local news articles and, via its Deep Mind company, music-generating software, an approach depriving the works of protection would be unwelcome news.
To date authorities in the USA and Australia have tended towards that approach while those in the UK, Ireland, Hong Kong, and New Zealand have instead favoured an approach in which the programmer is the author.
The UK’s Copyright Designs and Patents Act 1988 specifically addresses computer-generated works and contemplates the possibility of non-human authors. It provides that the author of a computer-generated work “shall be take to be the person by whom the arrangements necessary for [its] creation”.
Who makes the ‘necessary arrangements’?
Given the effort involved in developing the underlying programs and machine learning capabilities, it seems that a model which encourages investment in tech and rewards such efforts aligns with both the spirit of copyright law and the economic interests of parties pioneering AI as a creative medium. The matter is not clear-cut, however: the distinction blurs as we leave the era where the computer was a tool like a brush and canvas to be manipulated by the user and enter an age in which autonomous AI systems make independent creative decisions without human intervention. When should the reward be due to the creator of the program and when the user who applies the parameters for the creation?
For Caselles-Dupré of Obvious the vital actor is the “human in the loop, asking questions”, commenting that “If the artist is the one who holds the vision and wants to share the message, then that would be us.” How the courts will determine whose vision is behind each work produced remains an open question.
From portrait to landscape
One might regard Christie’s sale of Belamy as a pioneering move to recognise AI as a valid art medium or as more of a marketing stunt to appear cutting edge. Whichever view you take, how technological developments will affect the creative industries, intellectual property rights ownership, and the creative economy is undoubtedly an evolving landscape. AI is capable of churning out new works at an unparalleled rate and becoming increasingly adept at learning from and mimicking human creativity – a Japanese computer-generated novella reached the second round of a national literary competition in 2016. The implications of these trends, together with the impact of developments in blockchain technology for authenticity verification and rights protection, seem to be the beginnings of a major event horizon which will dawn on creatives, collectors and businesses alike.