AI under the hammer

Earlier this month a portrait of Alan Turing, widely considered to be the father of artificial intelligence, was sold at auction by Sotheby’s for £836,667. The sale price in itself is perhaps not unusual for sought after works of art but the fact that it was a portrait generated by AI certainly does make it more unusual.

The work, known as A.I. God, was created by Ai-Da Robot a “humanoid robot artist” or AI to you and I. Ai-Da gave interviews reflecting on the work and values embodied in it and has her own website giving further information about her influences and inspiration.

There is no doubt that the technical capacity of Ai-Da to create and reflect on her work is quite astounding but it does raise questions again about what art is and what people want from art. Do we want some human experience to permeate the art we see and enjoy or are we content for technology produce what we may hang or our walls or see in galleries?

For the time being, however, some human input is still required even for AI produced artworks in terms of the artworks the AI is trained on and the humans who instruct the AI to create its output. Perhaps the actions of the human artists instructing Ai-Da are, in fact, works of art in themselves.

Whatever your take from an artistic perspective, the creation of artworks by AI does raise some more prosaic questions about ownership and copyright.

In the scheme of legal disputes, relatively few copyright infringement cases make their way to court but copyright infringement claims relating to the use of original human authored works to train AI is steadily on the rise, particularly in the US, where various claims are being litigated by high profile authors whose work has been used to train AI models. However, the creation of works like A.I. God also raises interesting questions about ownership, particularly when the work may be considered culturally important or at least when it has a healthy price tag.

While there is continuing debate about ownership of AI output and regulation of AI, it remains the case in the UK that there is no overarching framework for regulation of AI. Instead, regulation is currently undertaken, to greater or lesser degrees, by regulatory bodies and organisations within different sectors of business and industry. The previous government’s proposal for a code addressing the copyright issues raised by generative AI was dropped earlier this year and the current law remains largely regulated by legislation passed before the advent of the smartphone.

That’s not to say, however, that the law fails to address these ownership issues and many of the basic principles about ownership of copyright still hold good. For example, the “author” (or creator) of artistic works will, subject to certain exemptions, be the owner of the copyright in the works they create. In the case of computer generated works, the author will be the person by whom the arrangements necessary for the creation of the work are undertaken. However, it is important to remember that employees who create works in the course of their employment will generally find that their employers own the copyright in their work and, where works are commissioned, the original author will remain the owner of the copyright (notwithstanding any specific commission) unless the copyright is formally assigned from creator to commissioner.

While there is much excitement and nervousness about the capabilities of AI, it is still important to consider some of the contractual basics about ownership and what can be done to regularise the legal position.