The question of whether artificial intelligence can be truly considered creative is something that has been asked and answered a thousand different ways. Since AI and machine learning first became a viable technology in the 21st century, society has questioned its ability to replicate the human brain’s ability to excel within ‘creative’ industries. Music, art, poetry and many more… What does it mean to be creative? How creative can a machine truly be in comparison to the human brain? Indeed, can AI truly ever be creative? It is a question that has a certain amount of fear and uncertainty around it. With many arguing that creativity is one of the last bastions of humanity and is a truly human function. Here at AI Music, as a technology company operating in an industry that is considered one of the most creative expressions – we have set out to provide our own interpretation of this difficult question. Read on to find out more… What does it mean to be creative? First, before you can answer the question of whether artificial intelligence can be creative, you must determine exactly what creativity means. Creativity is not a black and white. The dictionary definition of creativity is the use of imagination or original ideas to create something; inventiveness. Something that is remarkably open to interpretation. IBM Watson is probably one of the most well known AI technologies on the market. It has applications across advertising, health, finance and even film making. For John Smith, IBM Fellow and Manager of Multimedia and Vision at IBM Research, it is just as important to determine exactly what creativity means to answer this question too, ‘We still have to define what creativity means. We know some of the attributes have to do with finding something novel, unexpected, and yet useful.’ In 2016, IBM Watson successfully generated a movie trailer for the film Morgan that demonstrated its ability to take something that is traditionally considered a very human process – a process that can weeks – and turnover a movie trailer in a matter of minutes.
Now, the question is not here whether the AI can produce something more creative than the human brain. Only, can it manufacture something that is both creative and usually only within the remit of the human skillset. If we are basing the answer to this question simply based on these parameters, then yes, AI can be creative.
Should we differentiate between low levels and high levels of creativity? For Jeffrey Ng, Chief Scientific Officer at Founders Factory, the important question to both ask and answer for this topic is what levels of creativity are we considering to be valid?
For instance, how should we differentiate between an AI mechanism remixing a track and generating an entirely new track based on that, versus creating a track entirely from scratch? Many would argue that the former is, in fact, an expression of creativity – is it then that the origin of that creativity that makes it invalid? Borne from collecting thousands upon thousands of data on one particular genre, an AI can generate a rearrangement of a track or a variation of a song within that genre effortlessly. Does it matter where AI creativity comes from? The romantic, traditional and long-held idea of music creation is that is inspired by the world around us, experiences, and a unique appreciation for how certain instruments played at certain times can create a mood, a feeling, and a memory. This is not wrong. Yet, it is also true that music composition today can involve the very same learnings that machines themselves use to generate music. The music industry knows that for a jazz ensemble certain instruments should be used, for a pop track, generic lyrics work best, and for drum and bass there should be a certain tempo. AI is simply using those learnings that humans have collected over generations and applying that within a mathematical setting to produce tracks. AI Music CEO, Siavash Mahdavi on the question of AI creativity “Some people argue it’s the process and some people argue it's the output. If it’s just the output, then anything AI created with music is creative. If it’s the process, then we can go deeper into how we actually measure it. So, if I design something and it fits into my parameters of what we know as creativity i.e. someone drawing a cube vs. AI creating an abstract piece of music never heard before – how does this fit into the parameters of what we know to be creative? For instance, our 3D engine at Within enabled engineers to create exceptionally complicated aeroplane pieces. The 3D printed software enabled people to create things they previously hadn’t been able to create before. From a mathematical perspective, we were producing things twice the speed and half the weight with less material. Practitioners were able to use those tools and push the boundaries of design. This isn’t to say generative design tools have replaced engineers. On the contrary, it has just enabled more creativity within this field. At AI Music, the mission remains the same. Not to produce music out of thin air, but to enable users – even those with no experience in music production – to take charge of their own creativity and create their own music. That is how we define AI creativity” You can learn more by watching our CEO on the CogX panel regarding AI creativity here. Conclusion There is no right answer to whether artificial intelligence can be creative. Indeed, it is a question that is steeped in so much controversy that it becomes difficult to have an answer that does not threaten either side. Yet, it is clear that if one can clearly determine exactly what creativity is before asking this question, the answer becomes much more obvious. Get in contact with us if you would like to know more about some of our ground-breaking research at AI Music.