AI in Literary Criticism: A New Frontier or a Human Crisis?
As artificial intelligence continues to evolve at an unprecedented pace, its influence has extended far beyond industrial automation and data processing, now reaching into the deeply human realms of art and literary criticism. The integration of AI into creative and interpretive practices has sparked intense debate among scholars, artists, and technologists. At the heart of this discourse lies a fundamental question: Can machines truly understand, create, or critique art in a way that rivals human sensibility? In a recent article published in ACADEMICS, Liu Jianping, a professor at the School of Literature, Southwest University and visiting scholar at The Chinese University of Hong Kong, offers a comprehensive analysis of how AI is reshaping the landscape of literary and artistic criticism—and what it means for human agency, ethics, and aesthetic values.
Liu’s article, titled Literary Criticism: Artificial Intelligence and Its Challenges, presents a nuanced exploration of AI’s role not just as a tool, but as a potential subject and even a rival in the domain of artistic creation and interpretation. Drawing from philosophy, aesthetics, and media theory, Liu argues that while AI has demonstrated remarkable capabilities in mimicking artistic forms—from poetry to music to visual art—it fundamentally lacks the subjective consciousness, emotional depth, and historical situatedness that define authentic human creativity. More importantly, he warns of a growing trend toward the “dehumanization” of criticism, where data-driven metrics overshadow value-laden, context-sensitive judgments.
The piece is part of a larger national research project on “Literary Criticism in the Micro-era,” funded by China’s National Social Science Fund, underscoring the urgency with which academic communities are grappling with the implications of digital transformation in cultural discourse. With a DOI of 10.3969/j.issn.1002-1698.2021.05.007, the article has become a reference point in discussions about the ethical and epistemological boundaries of AI in the humanities.
Redefining Intelligence: Beyond the Myth of the Machine
One of the most persistent misconceptions about AI, according to Liu, is the notion that it represents a new form of autonomous, self-aware intelligence—what some futurists call “superintelligence.” Popular culture, from films like I, Robot to A.I. Artificial Intelligence, has fueled a romanticized vision of machines evolving beyond human control, capable of emotions, desires, and moral reasoning. But Liu cautions against such anthropomorphism. He emphasizes that current AI systems, no matter how sophisticated, are not sentient beings. They are tools—albeit highly advanced ones—designed to process information, recognize patterns, and simulate decision-making based on vast datasets.
The term “artificial intelligence” itself, Liu notes, dates back to 1950, when Alan Turing first proposed the idea of machines that could mimic human thought processes. The foundational concept was not to create life, but to replicate cognitive functions such as problem-solving, language comprehension, and pattern recognition. Today’s AI, powered by deep learning and neural networks, operates on the same principle: it learns from data, not from lived experience. Systems like AlphaGo, which famously defeated world champion Lee Sedol in 2016, do not “think” in the way humans do. Instead, they analyze millions of game positions per second, drawing on a database of past matches to determine optimal moves. Their “intelligence” is computational, not experiential.
Liu distinguishes between two conceptualizations of AI. The first is the non-biological, machine-based intelligence developed through programming and machine learning—what he refers to as “non-naturally born intelligent beings.” The second involves the integration of intelligent prosthetics into human bodies, such as neural implants or bionic organs, creating what some call “cyborgs.” While both forms blur the line between human and machine, Liu argues that only the former qualifies as true AI in the context of artistic production. The latter, he contends, is an extension of human capability, not a replacement.
The Illusion of Creative Autonomy
In recent years, AI has made headlines for its apparent forays into artistic creation. Microsoft’s AI “Xiaoice” published a book of poetry titled Sunlight Lost Behind the Glass Window in 2017—the first poetry collection entirely generated by a machine. Similarly, Tsinghua University’s AI system “Jiuge” composed classical Chinese poems so convincingly that audiences on a televised talent show mistook them for human work. In music, Georgia Tech’s Shimon robot can improvise jazz pieces after training on thousands of songs, while robotic ensembles have produced full electronic albums.
These achievements, impressive as they are, do not signify true creativity, Liu argues. Artistic creation, in the human sense, is not merely the recombination of existing forms or the optimization of stylistic patterns. It is an act of meaning-making rooted in personal experience, emotional resonance, and socio-historical context. When a poet writes of loss, love, or longing, they draw from a well of subjective feeling that no algorithm can replicate. AI, by contrast, operates within a closed system of data inputs and probabilistic outputs. It can generate text that resembles poetry, but it does not feel the poem. It does not mourn, rejoice, or reflect.
Liu illustrates this distinction by examining the internal logic of AI-generated art. Take Xiaoice’s poetry: while the verses contain emotionally charged vocabulary—words like “tears,” “loneliness,” “dreams”—these are not expressions of inner sentiment but statistical aggregations derived from user interactions across social media. The AI has no consciousness of sorrow or joy; it simply identifies linguistic patterns associated with emotional content and reproduces them. Similarly, painting robots do not conceptualize composition or theme. They begin at a corner of the canvas and fill in pixels based on pre-programmed algorithms, lacking the holistic vision of a human artist who imagines the final piece before the first stroke.
This mechanistic process, Liu asserts, reduces art to a series of computational steps—an “encoding and decoding” operation devoid of intentionality. True creativity, he insists, requires what Marx called “the actual transformation of the reflected world in the spiritual realm.” It is not the application of rules, but the transcendence of them. Artists do not merely assemble elements; they reimagine reality. AI, bound by its training data and algorithmic constraints, cannot achieve this leap.
The Crisis of Authorship and Originality
Another critical issue Liu addresses is the erosion of artistic originality in the age of AI. Because AI systems rely on vast databases of existing works, their output is inherently derivative. A robot trained on Van Gogh’s paintings can produce new images in his style, but these are variations, not innovations. Moreover, because the same algorithm can be replicated across multiple machines, AI-generated art lacks the uniqueness and irreproducibility that define authentic artistic expression.
In human art, no two works are ever identical, even when created by the same artist. Each piece carries the imprint of a specific moment, mood, and intention. AI, however, operates on reproducibility. Once a model is trained, it can generate endless variations of the same style, leading to a homogenization of aesthetic output. Liu cites the example of a calligraphy robot developed by Professor Xu Yangsheng’s team at the University of Hong Kong. The robot can write in multiple calligraphic styles with precision, but each stroke is determined by code, not by the fluctuating hand of a human scribe. The result may be technically flawless, but it lacks the “breath” of life—the subtle imperfections that convey presence and personality.
This raises profound questions about authorship. Who owns an AI-generated artwork? Is it the programmer, the trainer, the machine, or the public domain from which the training data was drawn? Liu points out that many AI-generated images are labeled as “generated” rather than “created,” acknowledging the absence of human authorial intent. This semantic distinction reflects a deeper ontological shift: when machines produce art, the very concept of the artist as a conscious, autonomous agent begins to dissolve.
AI as Critic: Precision Without Insight
If AI’s role as an artist is questionable, its function as a critic presents an even more complex dilemma. On the surface, AI appears well-suited for evaluative tasks. In 2018, China Central Television introduced “Xiaoke,” a music evaluation robot developed by the Institute of Automation at the Chinese Academy of Sciences. Xiaoke assessed singers on six dimensions—pitch accuracy, vocal range, tonality, rhythm, diction, and musicality—delivering scores with remarkable consistency. Unlike human judges, who may be swayed by emotion or bias, Xiaoke remained “the calmest judge in history,” as audiences dubbed it.
From a technical standpoint, AI excels in quantitative analysis. Its neural networks can detect minute variations in sound frequency—up to 30 times more sensitive than the human ear. It can analyze brushstroke density, color distribution, and compositional symmetry in visual art with precision unattainable by human observers. These capabilities offer valuable tools for art historians, conservators, and educators, enabling new forms of digital humanities research.
Yet Liu warns that such technical proficiency does not equate to aesthetic judgment. True criticism, he argues, is not merely the measurement of formal properties but the interpretation of meaning. It requires empathy, contextual knowledge, and philosophical reflection—qualities that AI fundamentally lacks. A human critic might interpret a painting as a response to political oppression, a personal crisis, or a spiritual awakening. An AI, by contrast, can only identify stylistic similarities or statistical anomalies.
Moreover, AI’s evaluative criteria are derived from historical data, reinforcing existing norms rather than challenging them. If trained on classical Western music, it will favor harmonic structures and tonal resolutions typical of that tradition, potentially marginalizing atonal or experimental works. In this way, AI risks becoming a conservative force in art, privileging conformity over innovation.
Liu also highlights the danger of “data centrism”—the belief that numerical metrics can fully capture the value of art. While big data can reveal trends in public taste or predict commercial success, it cannot explain why certain works move us deeply. The emotional resonance of a poem, the haunting quality of a melody, the sublime tension in a painting—these are not quantifiable. They emerge from the interplay of form, content, and lived experience, a synthesis that transcends algorithmic processing.
The Ethical and Philosophical Implications
Beyond technical limitations, Liu raises urgent ethical concerns about the role of AI in shaping cultural discourse. As AI-powered recommendation systems dominate platforms like YouTube, Spotify, and TikTok, they increasingly dictate what art is seen, heard, and valued. These systems operate on engagement metrics—clicks, likes, watch time—favoring content that is easily digestible, emotionally stimulating, or algorithmically optimized. The result, Liu warns, is a flattening of aesthetic diversity, where niche, challenging, or slow-burning works are buried beneath a flood of viral content.
This commodification of art, driven by machine logic, threatens the autonomy of both creators and audiences. Artists may feel pressured to conform to algorithmic preferences, sacrificing authenticity for visibility. Audiences, in turn, may lose the capacity for independent judgment, their tastes shaped by invisible recommendation engines. Liu draws a parallel to 19th-century industrial workers, who were reduced to appendages of the machine. Today, he suggests, cultural producers and consumers risk becoming “digital infants”—passive recipients of pre-digested content, stripped of critical agency.
The concentration of AI technology in the hands of a few powerful corporations exacerbates this risk. As Liu notes, while elite institutions and tech giants develop cutting-edge AI for art and criticism, the general public often accesses only basic, consumer-grade applications—smart speakers, fitness trackers, chatbots. This technological divide mirrors broader social inequalities, where access to cultural tools and creative platforms remains uneven.
Toward a Human-Centered Future
Despite these challenges, Liu does not advocate for the rejection of AI in the arts. On the contrary, he sees its potential as a powerful supplement to human creativity and criticism. When used ethically and transparently, AI can enhance research, expand access, and support artistic experimentation. The key, he argues, is to maintain a human-centered framework—one that prioritizes values, meaning, and subjectivity over efficiency, speed, and scalability.
Liu calls for a rethinking of AI development goals. Rather than pursuing ever-greater computational power, researchers should focus on integrating qualitative analysis into machine learning models. Can AI be taught to recognize irony, metaphor, or cultural nuance? Can it understand the difference between sincerity and pastiche? These are not just technical questions, but philosophical ones that require interdisciplinary collaboration between computer scientists, artists, and humanists.
He also emphasizes the importance of preserving human agency in the creative process. AI should serve as a collaborator, not a replacement. Just as the camera did not eliminate painting but transformed it, AI may not end human artistry but redefine it. The future, Liu suggests, lies not in a battle between man and machine, but in their symbiosis—a “cyborg” model where human intuition and machine precision coexist.
Ultimately, Liu’s article is a call for vigilance and reflection. As AI becomes more embedded in cultural life, we must ask not only what it can do, but what it should do. Art and criticism are not neutral domains; they are arenas of meaning, identity, and resistance. To cede them entirely to algorithmic logic is to risk losing what makes us human.
Liu Jianping, School of Literature, Southwest University, ACADEMICS, DOI: 10.3969/j.issn.1002-1698.2021.05.007