When creativity is linked to the production of art, literature and music, it feels natural to associate it with being human.
Famously, during the Second World War Winston Churchill responded to the suggestion of cutting arts funding to redirect money to the war effort with ‘Then what would we be fighting for?’ Somewhat less famously, there is no evidence that he ever made this comment. However, before the war he did say with reference to painting and sculpture ‘The arts are essential to any complete national life. The State owes it to itself to sustain and encourage them.’ Many view the arts as a central component of human endeavour.
What, then, are we to make of software that is capable of generating artworks, music, short stories and novels? Just as there are concerns about artificial intelligence making inroads into other intellectual activities such as coding, we know that AI is capable of producing apparently novel text, music and pictures. Some would say that it’s time to consider software capable of true creativity.
Throwing the Musical Dice
There are some existing ‘human’ artworks that have in practice been generated by algorithm. Composer John Cage, for instance, wrote a piece called Music of Changes where what is effectively a random selection process is used to assemble musical components into a piece. This is not even a modern idea. In the 18th century, games known as Musikalisches Würfelspiel (musical dice game) became popular in continental Europe with the best-known produced by Mozart’s publisher, who alleged that the great man had designed it.
However, for the most part it is felt that it’s the input of human intelligence that makes for creativity. When generative AI code produces a ‘new’ artwork or piece of writing or music it does not think in the same way as a person. The work is the product of taking influence from a whole range of existing (human-created) pieces and bringing that together to produce something different. Those who dismiss computer creativity point out that there is nothing truly original here.
Stealing Art
There is one flaw in this argument, though. A human artist, writer or composer does not exist in a vacuum. They are bound to have external influences – and some of the most creative members of the arts have recognised this. The poet and playwright T. S. Elliot wrote ‘Immature poets imitate; mature poets steal.’ Entertainingly, very similar comments have also been attributed to Stravinsky about musicians and Picasso about artists – it’s not clear who, in this case, was doing the stealing.
Perhaps, then, there is not too much difference between the influences of a human and the data used to train an AI creative. Could it be that we need to give in and admit that software with the right inputs has the potential to be creative?
For the future, there is now an increasing call for some form of licensing, where generative AI companies would pay those whose work was used as an input, much as the Authors’ Licensing and Collecting Society currently collects fees for copying published material. This seems reasonable, bearing in mind that the connection between the sources a generative AI is trained on and its output is far clearer than is the case with human inspiration.
Given how much the value of modern art is driven by name rather than quality of output, it’s entirely possible that human artists will not be displaced much by AI software. And much AI-generated art, writing and music is mediocre at best. But it will continue to improve – and it will become increasingly hard to say definitively that creativity is not involved.