by Madoc Wade
© 2024 Wade, M. All rights reserved
Citation: Wade, M. (2024) Is AI a Threat to Human Creativity? available at https://www.oxford-aiethics.ox.ac.uk/ai-threat-human-creativity
I argue that artificial intelligence is a significant and immediate threat to human creativity. Importantly, however, this threat is contingent on AI’s broader economic consequences, rather than on its ability to out-perform all human creatives. This essay has three stages. First, I will outline an operational definition of creativity to make clear exactly what is under threat. Second, I will distinguish between two different conceptual ways that AI could threaten human creativity. I call these the direct and instrumental threats. Finally, I will evaluate these threats independently. I conclude that the direct threat posed to human creativity by AI is negligible. The instrumental threat, however, is very real and already here.
Following Maria Kronfelder1 and Berys Gaut2 I propose that creativity is best understood as a propensity to create works which are both psychologically novel to the creator and appropriate. By psychological novelty, I mean work that is not an intentional copy of something else. The mere existence of a work does not intrinsically diminish the accomplishment of a similar one. For example, a hypothetical isolated painter who coincidentally paints Starry Night with no prior knowledge of Van Gogh may reasonably be called creative, despite the other painting already existing. As for appropriateness, I mean that the work must have value in a given problem space. Under this definition, even acts that do not produce value as we would commonly think of it could still be considered to be creative: an example might be a complex and elegantly planned murder. There is also a question of the intentionality and self-appraisal necessary for an act to be considered creative, what Gaut calls ‘flair’. Which criteria are necessary and sufficient for creativity is an interesting and important discussion, but one which I must leave to the side for this essay. The most important aspect of this definition is that creativity is a propensity for a type of action. Rather than being something that someone just has, creativity is something which exists only with respect to the actual process of creation. A person could still possess creativity even if they were restricted from actually creating, but only under certain external conditions, such as if they were imprisoned and so lacked access to the tools to actualise their vision.
It is now possible to distinguish between two ways that AI might threaten human creativity. First, the direct threat. As AI becomes increasingly advanced, its capacity for creative output might eclipse its human counterparts, and in doing so render their creativity irrelevant. The second possible threat is instrumental. In this case, the threat posed by AI is not one which emerges from the relation between human and artificial creatives directly, but rather from the way that AI will be integrated by existing profit-seeking economic structures in society. Due to the very low cost of production, AI will be seized upon for its ability to provide an inexpensive but adequate product compared to more expensive processes which require deep human creativity. The workers made economically redundant in this revolution will then no longer exhibit as much creativity as they had when they were paid to do so.
In short, the direct threat is to a person’s ability to create given some particular economic opportunity. It is the notion that AI will become so good at creation that human creation will be either outright impossible or rendered as such a pale reflection of its artificial counterpart that we stop caring for it. Instrumental threat, by contrast, involves the existence of those economic opportunities for creation, under the conjecture that the pursuit of creativity is directly tied to its economic feasibility. Regardless of its capabilities, AI does not pose a direct threat to our artistic creativity. For one thing, creativity has never been synonymous with our technical proficiency to write, draw, sing or engage in any other artistically creative activity. But this is not to suggest AI could never be creative, or that the product of its work will always fall short. Certainly, the current iterations of generative AI seem to lack the flair described by Gaut as an essential component of creativity – that intentionality and ability to critically evaluate their own work.3 But this may well be a short-term limitation4, and broader dismissal of the possibility of artificial creativity is symptomatic of a chauvinism which reflexively positions human creativity as metaphysically distinct.5 Creative AI will not make human creativity irrelevant because creativity has never been a rivalrous resource. Bach’s creativity did not exclude Vivaldi’s; the Sistine Chapel was not beaten to the punch by the Lascaux cave paintings. It is a truism that artists have always drawn on one another’s creativity as inspiration, and a fledgling artificial artist is no different. It will draw on its human contemporaries and predecessors as inspiration, and hopefully inspire work by and with them in turn. We may not appreciate the products of these creative processes as particularly insightful – although equally we may find meaning in them that we yet cannot imagine – but they will not preclude our appreciation of human artists.
Rather than representing a direct replacement, it is AI’s instrumental nature as a means of production that threatens our creativity. Companies offering AI ‘assistants’ to help with creative tasks have exploded over the last few years, which supporters purport will streamline, “augment”6, or “supercharge”7 the human productive process. While companies are careful to advertise their technology as only supplementing human creativity, the suggested applications paint a different picture. ChatGPT can write a press release or article in seconds; DALL-E can generate dozens of possible logo designs in that same time; ‘Appy Pie’ can provide a (rudimentary) website and a whole host of services claim to be ‘brainstorming AI’s’ – generating (or at least restructuring) ideas. These examples expose a clearly-defined domain in which AI encroaches on human creativity: creativity for profit. The creative processes involved in web design, copywriting, advertising and internal messaging are fundamentally instrumental. A website is not designed to be reflected on and marvelled at in the same way as a painting or musical composition. Its primary purpose is to direct and maintain attention on some other object; be it informational, commercial or social. These products – although certainly potentials for creative outlet - are not supported for the appreciation of creativity primarily, but rather for the creation, distribution and social reproduction of some other product. AI is already capable of producing work of a sufficient quality to meet the low bar of professional demand in many fields, and when coupled with the exceptionally low cost of such production, the financial choice compared to a human is becoming trivial.
But had we not established that creativity was non-rivalrous? Why can’t ChatGPT, Gemini or their cousins supplant the creative without threatening creativity itself? Recall that creativity is a propensity, not just a static talent. For the professional creative, their willingness to create – and therefore their creativity - is fundamentally dependent upon demand for their work. This is not quite the same as the example from earlier of a person who is imprisoned, and so is fundamentally incapable of creation. An unemployed person could still create – they still have the time in the day – but their choice as to whether they will or not is heavily influenced by their socio-economic reality and pressure to earn a living. Of course, not all creative works are produced for the sake of, or under the pressures of, external demand. Plenty of famous creatives died without even attempting to make money from their work; their creativity was actualised for their own reasons. Perhaps in a utopian economic system where survival was not dependent on economic output, the workers displaced in this revolution would continue creating just as they had been before. Perhaps their work could be more authentic, insightful and personal; more human. But we do not live in such a system. Creativity in itself may not be rivalrous, but the opportunities for creation that realise creativity are, and in a world in which AI replaces humans in those positions, human creativity will be diminished.
I have painted, in some respects, a bleak picture of the prospects for human creativity in the face of AI. Creative work in the modern day is largely sponsored, and thereby enabled, for economic purposes such as the advertising or design of some product or service. Due to its ability to produce work of a sufficient technical standard to achieve these instrumental goals at a lower cost, AI poses an existential threat to the livelihoods and thus work of an entire class of ‘professional creatives’. But this does not leave us without any human creativity at all; the future may be a world of casual weekend watercolorists, alongside a class of ‘elite’ artists who continue to command interest in their work, partly at least, because of its human origins. Human creativity is under threat from AI, but it still has a place in our future.
Bibliography
Kronfeldner, M. E. (2009). Creativity naturalized. The Philosophical Quarterly, 59(237), 577-592.
Gaut, B. (2010). The philosophy of creativity. Philosophy Compass, 5(12), 1034-1046.
Sæbø, Solve, and Helge Brovold. "On the stochastics of human and artificial creativity." arXiv preprint arXiv:2403.06996 (2024).
Hubert, K.F., Awa, K.N. & Zabelina, D.L. The current state of artificial intelligence generative language models is more creative than humans on divergent thinking tasks. Sci Rep 14, 3440 (2024)
Boden, Margaret A. (1998). Creativity and artificial intelligence. Artificial Intelligence 103 (1-2):347-356.
Eapen, T. T. et al. (2023, June 13). How Generative AI can augment human creativity. Harvard Business Review. https://hbr.org/2023/07/how-generative-ai-can-augment-human-creativity
Parkinson, M. (2023, November 9). Using AI to supercharge creativity & connection | IAB UK. https://www.iabuk.com/perspectives-trust-effectiveness/using-ai-supercharge-creativity-connection