Authentic vs. Automated: Where A.I. Fails
BY Makena Legault
With films such as The Matrix and 2001: A Space Odyssey being staples of our pop culture, it seems we have been envisioning a future where human intelligence is surpassed by that of its creations for generations- and that time might be closer than ever before. Recently, artificial intelligence, or A.I., has been making headlines, and for good reason. While it is by no means new (in fact, computer programs have been using it to play chess and checkers since the 1950s), its development and our reliance on it have been increasing at an astounding rate. Now, anyone with an internet connection has A.I. right at their fingertips, enabling them to converse with chatbots, produce full novels in seconds, or even generate artwork in the style of famous painters such as Van Gogh or Monet. Whereas once humankind and A.I. worked alongside each other, tools are becoming so advanced that many people are raising questions about whether or not they can replace human-produced work altogether. After all, in a world where artificial intelligence can supposedly replicate human skills in a quicker and cheaper manner, what is the value of continuing to rely on people to produce creative works? Though it may be tempting to transfer creative duties to these new inventions, a closer look at artificial intelligence shows that it may not be as helpful as it initially appears. Recent advances in A.I. technology, while undoubtedly impressive, are often created with seemingly little consideration of their potential impact and put the expression of human creativity in jeopardy.
One area where A.I. poses a significant dilemma is education; when services such as ChatGPT can spit out an entire paper faster than one can write their name, what incentive is there for students to write their own work? Unfortunately, one does not have to dive deep to discover the issues within AI-generated essays. Even on just a surface level, the quality of the writing produced by ChatGPT and similar tools tends to be subpar at best. It is all too common for A.I. to fail to properly cite information, to fall victim to logical fallacies, or to generate falsehoods presented as fact. Of course, even if a student using A.I. generated work was to correct these issues (many of which are so significant that amending them would require the student rewriting the paper themselves anyways), the final product still inevitably lacks much of the value of an original piece. Writer Stephen Marche explores this intrinsic worth in his article “The College Essay is Dead”. He expresses concern with A.I.’s invasion into the classroom, worrying that useful skills are left behind when artificial intelligence takes the reins. Marche argues that an auto-generated block of text does not hold the same weight as a genuine, human-produced essay, which he says “…[have] been the center of humanistic pedagogy for generations” (Marche 2). Essays are an invaluable tool, not only for teaching skills such as reading and writing but also for teaching us how to think critically and articulate our opinions. They are also an important reflection of the writer themself and give new insight into who they are as a person that readers cannot gain from the passionless text of a computer program. Artificial intelligence may be able to produce work that responds to the assignment at hand, but in the process, it abandons the development of important abilities such as these. The loss of this practice could undermine some of the key beliefs of education and leave learning a shallow and incomplete husk of its former self.
Of course, ChatGPT isn’t the only A.I. tool taking the world by storm. Through resources such as DALL-E 2 and Midjourney, words on a page can be transformed into full works of art with the click of a button. For people like Adam Gopnik, author of “What Can A.I. Art Teach Us About the Real Thing?”, this is an exciting prospect. While unimpressed by ChatGPT’s artificial prose, Gopnik sees potential in A.I.-generated images. He argues that the difference lies in the nature of visual art itself, writing that “to be persuasive, a text demands a point; in contrast, looking at pictures, we can be fascinated by atmospheres and uncertainties.” (Gopnik). Many written works rely on discovering a specific argument, intention, or style. However, much of the appeal of visual art depends on exploring and deriving meaning from its ambiguities, a quality that A.I. art retains. Being able to replicate visual art with greater success than text is also partly due to how A.I. operates. With text, A.I. learns to seek out and replicate linguistic patterns. The result is a string of words that separately align with the prompt but leave little to the imagination and thus fail to capture the author’s spirit. Without this spiritual component, it becomes much less likely for any deeper meaning to be found or for readers to form a connection to the work. However, art’s deeper meaning is not solely based on its spiritual component but also a visual component. Unlike spiritual meaning, visuals are something that humans and machines alike can create. A.I. art may not be able to replicate everything that human-produced art has to offer, but unlike A.I. text, it can replicate something. With this intact, many viewers of A.I. art can still single out aspects to find their own meaning within a piece, reproducing, to an extent, the experience of viewing traditional visual art.
Despite this, A.I. still faces one challenge that prevents it from truly replacing humans in the current moment: originality. Even Gopnik concedes that while A.I. works may occasionally succeed in eliciting emotion, “[it] is not a machine to draw the world. Instead, it proposes a recombinant approach to popular imagery as a means of making art” (Gopnik). Each image produced by artificial intelligence only exists because of the ideas and pictures that came before it. It cannot create anything entirely new because the data it would need to do so simply does not exist. Due to this, A.I. art challenges nothing and takes no risks. It is unable to convey anything truly meaningful because art is a form of human communication, a conversation between the artist and the viewer. When the so-called artist has nothing to say and is only spitting out pixels from an algorithm, a vital component of art is missing. Sure, people can still find ways to interpret a piece of A.I. art, but without a human artist behind the piece, this meaning ultimately rings hollow. Would a painting such as The Scream still be revered in society without the context of Edvard Munch’s struggle with mental illness? Would the surrealist art of Salvador Dalí still be memorable had it not broken the rules of what art should be? There is a reason that paintings of everyday things, such as sunflowers and ocean waves, have withstood the test of time while countless others have faded away. It is not merely because they look appealing but because they did something new and added to our cultural dialogue in one way or another. The inability of artificial intelligence to produce anything genuinely innovative results in art that changes little in our social landscape, proving how crucial human imagination is to society.
Unfortunately, the derivative nature of A.I.-generated images has already caused serious problems for creatives. The data programmers train artificial intelligence on is still the work of real people – work that is often used without the permission, or even knowledge of, the original creator. One such artist, Greg Rutkowski, was blindsided when he discovered his name had been used to prompt A.I. art nearly 300,000 times. When asked what his reaction was, Rutkowski said: “I felt really overwhelmed, and I’m still feeling really overwhelmed by all this …when it’s happening on this really large scale…you feel like anxious, definitely anxious for most of the time about your future” (Rutkowski). One of his major worries is the issue of intellectual property. A.I. art operates outside the jurisdiction of copyright laws, creating opportunities for others to generate and sell works that are, in essence, not their own. It may appear original, but at its core, it is still an amalgamation of stolen works. In cases like these, the original artists will receive no credit or compensation for their work, even if A.I. directly used it to create the product. A.I. has made it even easier for people to infringe upon an artist’s work, and as of now, there is not much they can do to fight back.
Intellectual property is far from the only anxiety artists like Rutkowski have about A.I.; many fear it could mean the loss of their jobs. At present, A.I. art still has enough flaws to dissuade companies from relying too heavily on it. Look closely at an A.I. image, and the imperfections will begin to emerge; hands will have too many fingers, glasses will appear to melt into faces, and background objects will blend into each other. However, as the technology continues to improve, these problems will be smoothed out, which many are concerned will result in companies taking advantage of the cheap and quick work of artificial intelligence rather than hiring real illustrators, photographers, or graphic designers. A.I. art may lack the same emotional impact as traditional art, but this is often not enough to dissuade businesses from choosing the path that they deem more economically sound. Even if a company does decide to hire human creatives, A.I. images may clog up the search results for an artist’s name, making it more difficult to find their portfolio. Rutkowski has already begun to experience this and worries “…potential clients will be turned off by all these fake Greg Rutkowski and won’t find his original work. He says that when he Googles his name, AI images already show up in the results” (Rutkowski). The damage A.I. art is already causing to human artists demonstrates the direct threat it poses to human creativity, even in its early stages.
In response to artificial intelligence’s challenges to creatives, Rutkowski and a group of fellow artists have chosen to sue. They claim that A.I. art is “…essentially a theft of their intellectual property. They weren’t asked if their art could be used to train this AI, and now this AI is specifically doing harm to them because it could replace them as working artists” (Rutkowski). Their goal is not necessarily to eradicate artificial intelligence but to put protections in place that ensure artists get their fair share of the wealth it may create. While winning the lawsuit would undoubtedly be a significant step forward for artists, it is uncertain how far-reaching any resulting legislation would be, especially when A.I. faces so little regulation as it is. There is also ambiguity as to which fields of creativity the decision would cover. A.I. affects more than just visual artists, with the replacement of humans by artificial intelligence becoming a source of tension among writers and actors as well. Regardless of the outcome of the case, it is probable that creatives, who are already largely overworked and underpaid, will need to learn to adapt to an entirely new and quickly evolving set of obstacles to keep doing what they love.
The issue of unoriginality also exists within text generated by artificial intelligence, though it raises a different concern. Some, like Noam Chomsky, say that this shortcoming of A.I. prevents it from being capable of the true intelligence many claim it to have. In “The False Promise of ChatGPT”, Chomsky dissects the supposed intellect of A.I., arguing that it is not all it appears to be. Intelligence, Chomsky claims, is about more than describing what is or predicting what will be. It also requires knowing “…what is not the case and what could and could not be the case. Those are the ingredients of explanation, the mark of true intelligence” (Chomsky). The inability to create new ideas is central to true knowledge, and no amount of facts that a machine may be able to recite will be able to make up for that. True intelligence also requires an understanding of ethics and morality, something that, by design, A.I. is unable to comprehend. When prompted, chatbots refuse to take even the most milquetoast ethical stances out of the desire of their creators to avoid controversy. Many of today’s most pressing questions are far too complex to avoid an opinionated solution. These problems require both new ideas and moral stances to be resolved, and when artificial intelligence lacks the capabilities to muster up either, it is unlikely that it will be able to be the all-powerful tool many portray it as.
Relying too heavily on something that does not possess true intelligence has the potential to hinder societal development. One of Chomsky’s major concerns is that A.I. will “…degrade our science and debase our ethics by incorporating into our technology a fundamentally flawed conception of language and knowledge” (Chomsky). The blindspots of A.I. are great enough to prevent it from solving most problems, and attempting to use it to do so is an exercise in futility. The solutions it provides for philosophical questions are meaningless without a grasp of moral concepts; applying solutions without consideration of how they could impact people is reckless and unethical. Additionally, A.I. serves little use in solving analytical problems without the ability to distinguish fact from fiction. Its tendency to “either overgenerate (producing both truths and falsehoods, endorsing ethical and unethical decisions alike) or undergenerate (exhibiting noncommitment to any decisions and indifference to consequences)” (Chomsky) renders it useless in spurring any sort of change. Developers are spending an extraordinary amount of time and resources in an attempt to complete an impossible task: giving these machines inherently human skills. It is hard not to consider the progress that might have occurred had these efforts been directed toward different causes.
Rest assured, as it currently stands, humanity is unlikely to fall at the hands of HAL 9000 or Agent Smith- but that doesn’t mean we don’t risk losing anything in the process. Sure, robots will not annihilate the human race anytime soon, but isn’t human creativity worth protecting anyway? Enacting legislation that restricts the use of A.I. and prevents it from replacing the work of humans would not be a complete fix, but it would certainly be a step in the right direction, one that ensures that artists like Greg Rutkowski can continue doing what they love and filling the world with meaningful art that only the human soul can create. It might not be the cheapest or quickest option, but the best compositions cannot be produced by a machine running a set of codes. Authentic human creativity is irreplaceable; it both comes from and elicits feelings that will forever be out of grasp from any algorithm that seeks to understand them.
Works Cited
Chomsky, Noam. “The False Promise of ChatGPT.” The New York Times, 8 Mar. 2023, https://www.nytimes.com/2023/03/08/opinion/noam-chomsky-chatgpt-ai.html.
Gopnik, Adam. “What Can A.I. Art Teach Us About the Real Thing?” The New Yorker, 1 Mar. 2023, https://www.newyorker.com/culture/cultural-comment/what-can-ai-art-teach-us-about-the-real-thing.
Knutson, Ryan, et al. “When AI Comes for Your Art.” The Wall Street Journal, Dow Jones & Company, 7 Mar. 2023, https://www.wsj.com/podcasts/the-journal/when-ai-comes-for-your-art/6873601d-dc2f-496b-ac14-2c0829e7b068. Accessed 4 May 2023.
Marche, Stephen. “The College Essay Is Dead.” The Atlantic, The Atlantic Monthly Group, 6 Dec. 2022, https://www.theatlantic.com/technology/archive/2022/12/chatgpt-ai-writing-college-student-essays/672371/.
About the Author
Makena Legault is a sophomore attending Fordham College at Rose Hill. She plans to major in New Media & Digital Design with a concentration in art, text, and design. In her spare time, Makena enjoys photography, listening to music, and riding the 4 train down to the Staten Island Ferry.