While the result of an AI generated art may superficially look similar to human-generated art, the process that results in its production is in no way similar. Study and imitation of an accumulated data set is not the same as individualistic creation
Marvel star Scarlett Johansson has accused ChatGPT of ripping off her voice. (Image credit: Instagram @scarlettjohanssonworld)
New Delhi: It seems that AI research company OpenAI and its CEO Sam Altman, creators of popular AI-based language model ChatGPT, cannot stay out of trouble for long. From boardroom troubles becoming public scandals to accusations of data theft and immoralities, troubles for the company seem endless.
The recent controversy being American actress Scarlett Johansson being “shocked, angered” by OpenAI’s ChatGPT voice that sounds like her.
What is the controversy
In a statement released by American actress Scarlett Johansson, she said how OpenAI CEO Sam Altman in September had asked her to voice ChatGPT’s text-to-speech product because he found her voice “comforting”.
She reportedly declined the offer. Nine months later, one option on GPT’s new voice product sounded so much like Johansson that she was “shocked, angered and in disbelief that Mr. Altman would pursue a voice that sounded so eerily similar to mine that my closest friends and news outlets could not tell the difference.”
OpenAI and Sam Altman have come out and denied the accusations, saying that the voice of their product Sky is not that of Johansson.
The company has still said that it is pausing the product for now and that the voice of their product “was never intended to resemble hers (Johansson)”.
The controversy though is not isolated to this specific instance but points to a larger trend. From copyright issues to the extent of duplicity and its morality, the latest technology of Artificial Intelligence seems to have raised questions on the limits of human intelligence, creativity, adaptability and transcendence, like never before.
How does art generation by AI work?
The use of artificial intelligence tools in creating content and art pieces of all kinds has seen a major burst. AI today is being used to create everything from poetry, visual art, videos to music.
This has led many artists to allege that AI apps are essentially ‘stealing’ their content. This has led many campaigns and petitions to instate more stricter copyright laws for AI-generated art.
AI art is also called generative art. It uses advanced machine learning models and algorithms to create art. It does so by using systems that study innumerable preexisting data relating to what the system is tasked with creating. Like most machine learning software, they identify and then replicate patterns in earlier data. This preexisting data is human created and it is here the ethical question of copyright claims and theoretical questions of authenticity in art arise.
A good example that sums up the whole issue is of Hollie Mengert, a Disney illustrator. A mechanical engineering student in Canada downloaded 32 of Mengert’s art pieces and using machine learning trained a model to reproduce her style. Shocked, Mengert was quoted saying, “For me, personally, it feels like someone’s taking work that I’ve done, you know, things that I’ve learned — I’ve been a working artist since I graduated art school in 2011 — and is using it to create art that that [sic] I didn’t consent to and didn’t give permission for.”
A question of authenticity and innovation?
Seemingly the main question raised by the barrage of recent issues relating to the use of AI, and its intersection with art, is how authentic such art is and can it be compared to that created by human minds.
The aims of art have been debated since millennia but its importance to human civilisation is generally agreed upon. Is art created just for its own sake, for reasons of self-expression, or is it for practical purposes, be it social or political.
Despite these questions, the creation of art by the human mind has been a constant process. It may have seen different phases in the degrees of innovation or excellence, from wall paintings by prehistoric humans to Renaissance frescoes, the need for our species to express emotions and ideas has been ever-present.
The workings of the process of transforming the inner self through a communicable medium is still not well-understood. However the quality of art seems to often be judged in the context of the effect it has on the recipient. This triggering of a response in a viewer is also seen in terms of how ‘authentic’ a piece of art is.
It is often argued that only a piece which is ‘authentic’, coming from a distillation of an artist’s accumulated experience and expressed in a state of truthfulness, is effective. This process is intrinsically and inexplicably human. It involves levels of perception, cognition and consciousness which are only known to be possessed by human beings.
This brings machine generated art in a heavy competition with that generated by humans. While the result of an AI generated art may superficially look similar to a human one, the process that results in its production is in no way similar. Study of an accumulated data set is not the same as individualistic creation.
Granted that artists too create from their accumulated experiences and a perceptual understanding of the world, ‘authentic’ art though often transforms this disparate inputs into something that transcends it. When it comes to creation of authentic ‘art’, the whole is way great than the sum of its part.
This machine learning cannot do yet and despite heady proclamations of technocrats, there is no guarantee if it ever could. AI-generated art, by its very nature, cannot exceed its input data. This lack of its capability to transcend the initial tools it has into something larger is what prohibits it from producing an authenticity that can match human art.
While questions of copyright are just, what is important to note is that a major distinction between the two modes of art creation exists and it may take long before AI-generated art can reach levels of authenticity and excellence shown by that created by man. Nonetheless till that happens, if it will, rather than being fearful of it, we should accommodate it and integrate its usage in how we operate and create art.
Technological innovations have always been inevitable and have historically changed art. This though is not always to art’s detriment. Similar should be our outlook with AI and we should use it to sharpen our artistic instincts and innovate without letting go of what truly makes our art human. This is the way forward and has been so always.
(Disclaimer: The views expressed in this article are those of the author alone. The opinions and facts in this article do not represent the stand of News9.)
