Ted Chiang, one of the foremost science fiction writers active today – he wrote “Story of Your Life”, the short story on which the film Arrival was based – recently published an article on the possibility, or rather, the impossibility, of AI creating genuine, or at least interesting, art.

His argument is premised on a perceived disparity: on the one hand, art requires making countless deliberate choices.  In turn, generative AI programs operate on the basis of human prompts which themselves define a set of choices. However, and herein would lie the disparity, the program has to make all the other choices not specified in the prompt. 

In the case of text generation, AI is known to do so in two ways: either by using an average of the choices made by other writers –which yields nondescript results, or by mimicking another writer’s style –producing highly derivative work. Even though the choices made by painters and photographers may be harder to quantify, the principle, Chiang argues, also applies to the visual arts.

Chiang thus takes issue with the promotional claim that AI can help artists to ‘unleash their creativity’ by minimizing the gap between an artist’s primary vision and their finished work. He claims that, precisely because the creative process requires making nuanced choices at every level and at every step of the process, the distinction between ‘inspiration’ and ‘perspiration’ is very faint. While it is not inconceivable for meaningful works of art to be produced on AI platforms (and some are), the sophistication of the underlying prompts renders the process as strenuous and time consuming as any medium. Furthermore, such use requires subverting the software’s purpose and mass market appeal, which is to render effort negligible. 

Chiang points out that even mundane tasks such as writing an email or a student report involve a loss when automated. As illustrated by the public dismay over a child’s fan mail to an olympic athlete written by Google’s Gemini, what makes writing meaningful is not so much its eloquence or grammatical soundness, but the fact that it refers back to a specific person who bore an intention to communicate. 

Through an array of thought-provoking anecdotes and analogies, such as pairing a software’s appearance of understanding with the resemblance to predator’s eyes in the spots on butterfly wings, Chiang arrives at identifying labor, intention and an anchoring in multigenerational shared experience as irreducible traits of meaning, and, by extension, meaningful art.