From ChatGPT to Sora: Innovation in Literature During the AI Era

2024 marks the centenary of the birth of the two masters of martial arts literature, Jin Yong and Liang Yusheng. Although the masters have long departed, I still remember that whenever martial arts secrets or extraordinary skills appeared in their novels, it would invariably lead to a bloody storm in the martial arts world. This reminds me of the recent global surge of artificial intelligence (AI) technology. From the sudden emergence of ChatGPT to the stunning debut of Sora, the “hot and spicy” AI is no less impactful than the Heavenly Sword and the Dragon-slaying Saber from Jin Yong’s works, stirring up a pool of spring water worldwide upon its appearance, with some panicking, some cheering, and some observing. Now, the video generation model Sora launched by OpenAI is increasingly embedded in people’s lives, and discussions surrounding the relationship between AI, humans, and literary creation are still in full swing.

The Joys and Concerns Brought by AI Development

What impact do AI technologies like ChatGPT and Sora have on literary creation? Will “AI creation” replace “human creation”? Currently, similar “threat theories,” “replacement theories,” and “control theories” are not uncommon, causing anxiety among some literary workers. Domestic scholars point out that the emergence of Sora will have a strong impact on ten major industries, including short videos, film, gaming, and advertising, putting relevant professionals at risk of unemployment. Meanwhile, abroad, more and more authors are joining the ranks opposing AI technology. In July 2023, the UK Writers’ Association released the “AI Survey Report,” revealing that 65% of respondents were very concerned about reduced writing income due to AI, and 61% were worried about job replacement. However, some maintain an optimistic stance. In March 2024, two Nobel laureates, Mo Yan and Gurna, held a dialogue in Beijing, where Mo Yan humorously stated that technological progress often gives literature new wings, and the emergence of AI will not pose a significant crisis for literature. He and Gurna will not be unemployed because “the uniquely individual image thinking of writers cannot be replaced by AI.” AI scientists Kenneth Stanley and Joel Lehman believe that Sora is not a monster; “If you are a painter but cannot invent a canvas, Sora is that canvas. It has been invented, and your talent can be displayed. Accessible technology can help you layer your ideas but will not replace you in creating entirely new ideas.” Indeed, as AI progresses, some rejoice while others worry.

So, for literary creation, is the leap of AI technology a “Pandora’s box” or a terminator, or is it a blessing or savior for humanity? In two recent articles published in the “Literary Newspaper,” titled “Can Sora Lead Us into a New Era of ‘Human-Machine Dance’?” and “From ‘Heart Literature’ to ‘Chip Literature’—Algorithmic Emotions, Ghost Images, and the Hidden Worries of ‘Chip’ Rules,” the authors present different views. Ma Bowen believes that the emergence of Sora presents more opportunities than challenges for online literature, suggesting that both should empower each other and develop together, leading us into a new era of “human-machine dance.” Zhou Zhiqiang emphasizes that the game rules of text-to-video may not only change literature but also change ourselves, stating, “The most concerning thing is that more and more people will like Sora’s images, immersing themselves in the bizarre and fantastical game images brought by Sora,” thereby distancing themselves from complex and authentic life experiences. Integrating the viewpoints of both scholars, I believe that literary creation in the AI era will undoubtedly undergo many changes, with both opportunities and challenges coexisting, but primarily it is an unprecedented opportunity. This opportunity is mainly reflected in the continuous iteration and enhancement of AI technology, leading to an ever-expanding boundary of AI creation, from early news reporting, current affairs commentary, and official documents to the current novels, poetry, prose, scripts, short videos, and various literary works, with AI’s presence almost ubiquitous.

This has profound implications for human literary creation. On the one hand, the threshold for creation in the AI era is lowered; anyone can participate if they are willing. The performance of AI in the literary field has garnered widespread attention, and more and more people are getting involved, objectively facilitating the prosperous development of literature. On the other hand, as a form of “new productive force,” AI technology can greatly enhance the efficiency of literary creation. In the martial arts world of Jin Yong and Liang Yusheng, “In the world of martial arts, speed is invincible.” Similarly, the greatest advantage of AI creation lies in speed. Reports indicate that a team led by Wang Feng, the dean of the School of Communication at East China Normal University, recently completed a one-million-word novel titled “The Apostle of Destiny” using a domestic large model in just one and a half months, whereas a diligent online writer might take a year and a half to produce a novel of similar length. In February 2024, South Korea’s Snowfox Books published a book titled “45 Ways to Find Life Goals,” in which the entire process—from generating English content using ChatGPT to translating it into Korean, proofreading, and cover design—was completed entirely by AI, taking just seven hours for content generation, two hours for translation, and less than a week for the entire publishing process, which is incomparable to traditional publishing. The Sora text-to-video model has even reached lightning-fast speeds, with remarkably realistic image quality. Although still in the experimental stage, if AI creation can achieve both quantity and quality in the future, it may give rise to new forms of literature distinct from traditional forms, such as “AI literature,” “AI film,” “AI music,” and “AI art,” similar to the emergence of “online literature” in the past.

The Three Major Challenges on the Path of ‘AI Writing’

The rapid advancement of AI technology has evidently brought changes to literary creation, but humans need not be overly concerned or anxious, as AI creation faces numerous problems and challenges if it aims to replace human creation.

Challenge one: quality of works. The key to the development of the literary cause lies in creating classic works. While classic works do not have fixed standards, the premise is that readers find them good. If AI is regarded as a writer or artist, the emergence of “AI literature,” “AI film,” “AI music,” and “AI art” as important new forms of literature requires a substantial accumulation of good works, which current AI creations evidently lack. For example, regarding the aforementioned AI work “The Apostle of Destiny,” Professor Wang Feng admitted: “It is still at the downstream level of online novels; although this novel appears to have some literary quality in certain plot segments, these fragments still need human assistance to be pieced together into a complete and coherent story.” These issues may potentially be resolved through deep learning, but will a powerful AI necessarily produce good works? Not necessarily. Taking Sora as an example, its text-to-video capability is its biggest gimmick; beyond that, it still has a long way to go to mature into a film product. Similar to directors who gain fame through unique skills, Sora merely provides faster video works for a diversified market. Just as China’s four great classical novels or the martial arts novels of Jin Yong and Liang Yusheng have been remade continuously with the advancement of film production technology, the quality of different versions varies, with older versions often recognized as the most classic. British poet Coleridge once said: “Everywhere there is water, yet not a drop to drink.” The literary creation of the AI era may fall into such a situation, where a vast number of works are mixed with poor quality, and truly high-quality literary works will always remain scarce. Moreover, despite the strong learning capabilities of ChatGPT and Sora, and their rapid content generation, the issue of overproduction must be considered. Human beings have a limited capacity to read and appreciate texts, videos, etc., during their lifetime; even if brain-computer interface technology becomes a reality, its receiving capacity has an upper limit.

Challenge two: innovative thinking. Innovation is the soul of literary creation and the driving force behind excellent works. AI creation is always based on human creation and can at most be considered a mover and assembler of existing resources, rather than a true original creator or thinker. A regulated AI respects only algorithms and programs, making it a “one-track mind” that does not deviate, and thus innovation becomes impossible to discuss. According to critic Xie Youshun, good literature is an accidental state of language and humanity. “If ChatGPT writes, it certainly cannot produce the sentence ‘One is a jujube tree, and the other is also a jujube tree’ from Lu Xun, as it clearly does not conform to language norms; nor can it produce Li Bai’s ‘The bright moonlight before my bed, I suspect it is frost on the ground. I raise my head to gaze at the bright moon, and lower my head to think of my hometown,’ because the poem repeats ‘bright moon’ twice; nor can it write ‘I looked at him once, then looked again, and I cried’ as Wang Zengqi did in his tribute to Shen Congwen—machines cannot write in such a manner.” Indeed, AI writing always follows conventional logic, while writer Wang Anyi believes that our novels and lives often do not follow conventional logic. It is precisely for this reason that writer Yu Hua asserts: “If ChatGPT were to write a novel, it would probably produce mediocre rather than unique novels; perhaps it could write seemingly perfect prose, but essentially it remains mediocre.” Mediocre literature does not last long; without innovation and imagination, how many readers can such works attract?

Challenge three: emotional resonance. Although human joys and sorrows are not universal, excellent literary works can enable different people to find emotional resonance. The reason for this is that the author invests genuine feelings, as Cao Xueqin stated: “A paper filled with absurd words, a handful of bitter tears. People say the author is infatuated; who understands the meaning within?” Good literary works “come from the mundane world and reach the soul.” AI itself lacks consciousness and cannot experience the surging waves of emotions, making it difficult to express heartfelt joys and sorrows in its works, leading to a situation where the virtual replaces the real, and formulas are mistaken for emotions. As Walter Benjamin said in “The Work of Art in the Age of Mechanical Reproduction”: “Even the most perfect reproduction of an artwork lacks one element: its real existence in time and space, and its unique presence in its unique location. For artworks, the core is their authenticity, which is even more solid in natural objects.” In other words, the genuine emotions that touch people’s hearts are the core of literary works, and since AI lacks the human experience of joy and sorrow, it is naturally difficult to create works that resonate with us.

Reflecting on the Foundations of ‘Human Writing’ from ‘AI Writing’

The quality of works, innovative thinking, and emotional resonance are both the chasms that AI writing needs to overcome and the important prerequisites for human literary creation to maintain its value. The prospects of AI creation depend largely on its technological development level, and in the foreseeable future, it is unlikely to replace human creation. As for the prospects of human literary creation, it entirely depends on humanity itself. Therefore, the enemy of literary creation is not AI, but humanity itself; what literary creators need to surpass is not AI, but humanity itself. How to surpass? Only through innovation. How to innovate? By stepping out of comfort zones, learning new techniques, gaining new perspectives, and creating high-quality works. Even if AI ultimately evolves into silicon-based life, the relationship between carbon-based humans and silicon-based life is not necessarily adversarial; instead, they can coexist harmoniously, advancing together to create works that better meet the spiritual needs of future humanity. Thus, “human-machine integration” could be a promising path for literary innovation in the AI era. In fact, countless pioneers have already embarked on this road of innovation.

As I write to this point, I suddenly wonder, what does AI think about “the impact of AI on current literary creation”? Will literature be replaced by AI? Surprisingly, someone has already posed this question to ChatGPT; let us take a look at its answer:

I believe that literature will not be completely replaced by AI. Although AI has made certain progress in generating text, literary works are not merely combinations of language; they also encompass emotion, imagination, and human experience. Literary works typically reflect the author’s unique thoughts, feelings, and aesthetic views, which are uniquely human traits that are difficult for AI to simulate or replace. While AI can assist in creation and provide inspiration, in creating literary works with depth and emotional resonance, human creativity and emotional understanding remain irreplaceable.

I must say that ChatGPT’s answer surprised and embarrassed me. I was surprised that its understanding of this issue coincided with mine, and in some respects, it was even more comprehensive; I was embarrassed that if its answer were considered a piece of writing, the aforementioned content flows smoothly, logically clear, and structurally complete, surpassing the writing standards of many individuals. However, I quickly felt relieved. Indeed, the uniqueness of humans in literary creation signifies its inexhaustible vitality. Given this, why should we trouble ourselves unnecessarily? Let us embrace AI passionately, as if welcoming a friend with shared ideals, to explore the vast sea of literary creation together.

[The author is an associate researcher at the Center for Humanities at Southern University of Science and Technology. This article is a stage achievement of the Guangdong Provincial Philosophy and Social Science Planning 2023 annual discipline co-construction project “Research on the Construction Path of the Literary Community in the Guangdong-Hong Kong-Macao Greater Bay Area” (Approval No. GD23XZW18)].

Content Source: “Literary Newspaper” April 17, 2024, Page 2

WeChat Editor: Wang Hongze (Intern)

Second Review: Xu Wanni

Third Review: Li Xiaocheng