What do we lose when we lose the ability to write? The rise of AI writing applications has led to a lot of understandable consternation, especially among teachers of writing. ChatGPT is already becoming a tool for rampant plagiarism, robbing students of the chance to demonstrate their understanding and develop critical thinking skills, while forcing teachers to overhaul their homework assignments and exams on the fly. But, as a university creative writing and literature instructor, I can see there’s something equally important students may be robbed of: the capacity to empathize and connect with others and with oneself that writing practice, at its best, can foster.

The concern that new technology is curtailing our capacity for empathy is nothing new. Sherry Turkle has written extensively about how technologies in the last two decades have diminished opportunities for face-to-face communication. The smartphone has become a perpetual social crutch, providing an easy escape from the mundane, uncomfortable, or difficult moments of face-to-face interaction. As Turkle explains, these technologies have resulted in less capacity to connect and empathize with others since we are spending less time literally seeing them: “We are seeing, with our children, in our romantic relationships, in our educational system, at work, that we are not paying attention to each other.”

The same is true with the move to virtual classroom settings, accelerated by the pandemic. While face-to-face interactions over Zoom still enable, to an extent, students and teachers to read non-verbal clues and body language—markers enabling empathy and connection—the online classroom remains a far cry from the actual classroom. I’ll admit that when I took certain graduate courses online, I was sometimes surfing the internet or responding to an unrelated email instead of giving the teacher or fellow students my undivided attention. Even when you are paying attention, it can be exhausting and alienating to stare at a screen of tiny faces while monitoring one’s own expressions in ways that feel unnatural.

AI poses analogous barriers to human connection and empathy, though it does so in a very different way. Technology critic L.M. Sacasas has written on what is lost when we contemplate an AI-generated image. He admits that such images can be delightful, alluring, and impressive. However, in his experience, the difference between gazing at an AI-generated image and the work of a human artist (such as Rembrandt or Bruegel) is that the former doesn’t allow for sustained engagement. It offers only surface-level appeal, depriving the viewer of the depth and insight that can emerge during prolonged contemplation. AI artworks present “surfaces that may arouse a desire to know more deeply but which do not have the depth to satisfy that desire.”

Similarly, when you read something written by a flesh-and-blood person, you’re experiencing the real emotions and thoughts of a being with conscious experiences of the world like your own. By reading it, you are able to retrace those thoughts and feelings. In fact, there is even evidence that a writer’s and reader’s brain waves become synced in a literal type of “mind melding” through shared language.

This is not the case with AI-generated writing. Let me give you an example. I asked ChatGPT to write a short fictional paragraph about a man named Leo being fired:

Time stood still for Leo as those words echoed in his ears, the full weight of their meaning crashing down upon him. His dreams, ambitions, and years of dedication to the company suddenly felt like an ethereal mirage slipping through his fingers. The room blurred as his mind raced to process the sudden turn of events. Every sound became muffled, as if the world had collectively pressed pause. He felt a kaleidoscope of emotions erupt within him: shock, disbelief, anger, and even a pang of humiliation. The room became a small prison, trapping him in a moment he wished would disappear.

This resembles human writing in many respects: it’s grammatical; it tells a coherent story; it uses imagery. Yet there is clearly something off—details that come close to human expression but remain not quite convincing.

This is what’s sometimes referred to as the “uncanny valley” phenomenon. It is the result of the way AI language models work, not by generating and expressing ideas but by producing amalgamations of terms, assembled in sequence, to emulate the movement of a story. The only “connection” this language conveys is to an inanimate technological process. The upshot is images that don’t quite make sense (“an ethereal mirage slipping through his fingers”); questionable metaphors (what do muffled sounds have to do with the world “collectively” pressing pause?); and a hodge-podge of clichés that add up to less than the sum of their parts.

The poor quality of AI writing is something different than that of bad high school or undergraduate writing. The latter comes from confused ideas and undeveloped powers of expression, but, for the most part, there is still something for a writing teacher to latch onto and try to tease out: evidence of thinking and the struggle to express oneself. With AI, it’s all surface; the errors and infelicities point not to deeper ideas struggling for expression, but an inhuman process incapable of depth.

In this vein, the historian Dan Cohen asks,

An AI text generator very well might spin a decent tale about a monomaniacal hunt for a white whale, perhaps even with copious Biblical references, given the right additional nudges, but would that work ever have the strange richness produced by a human writer familiar with the actual manual labor of whaling, and who is able to find layers of meaning in those seemingly mundane processes?

What we will lose by relying on ChatGPT for literary works, whether in the classroom or beyond, is the communal experience of knowing and forming a transient union with another intelligence. We’ll lose opportunities to break out of our own subjective experience—our own naturally myopic and self-absorbed view of reality—and relate to another person, plumb and feel another’s emotions, imagine alongside them their dreams, experience a sense of kinship and belonging that unites us as human beings.

We might (eventually) be entertained by AI-generated TV shows or charmed by AI-generated poems, but we will be deprived of one of the joys of inhabiting the mind of a human creator and the sense of connection that emerges from such an experience. Just as when we gaze at a smartphone and ignore our neighbor, we will miss out on an opportunity for connection. As a result, our worlds and lives will become smaller, less connected, and more isolated. Without real writing, we will become strangers to each other, losing our sense of shared humanity.

In his essay, Sacasas draws on Hannah Arendt’s claim that art is a bulwark against loneliness because it fosters communion between those who engage with it as well as create it. Put another way, when we engage with the writing or art created by another person, we are engaging with an intelligence that knows what it’s like to be a human being and shares our own experience of reality. This, as Tolstoy writes in his book What Is Art?, is what makes art valuable and meaningful to both the individuals experiencing artworks and the larger culture to which they contribute. Tolstoy writes,

The more individual the feeling transmitted [through the artwork by the artist] the more strongly does it act on the receiver; the more individual the state of soul into which he is transferred, the more pleasure does the receiver obtain, and therefore the more readily and strongly does he join in it.

In the classroom, the rise of ChatGPT presents students and teachers yet another barrier to connection and empathy, on top of the smartphone and Zoom. Part of the joy of reading my students’ work is that I am granted passage into their interior lives. In some meager way, I get to experience what it’s like to see the world as they do, to think as they do, to learn from their distinct perspectives. By entering their world through their writing, I’m better able to nurture their inchoate insights into something, hopefully, more cohesive and compelling. I can challenge them to develop their ideas and, in so doing, to develop deeper senses of themselves. If AI writing comes to replace student writing, these opportunities for students to process and articulate their experiences and for teachers to enter into and impact students’ worlds will disappear. This is a terrible loss.

Still, we might ask: What about ChatGPT’s use when it comes to non-creative writing, such as research papers, critical essays, or written exams? It’s true that a research essay does not typically connect us with fundamental human experiences in the same direct way more creative writing can. However, there is still a great deal of creativity required in the case of research writing: making decisions around voice and style, the skillful marshaling of evidence in a fluid manner, the connecting of ideas and insights that reveal the writer’s perception of and beliefs about the world. It may not offer the same degree of human connection and empathy that literary works do, but it nevertheless invites the reader to inhabit and understand the mind of the writer.

Whether a student is tasked to write a short critical reflection or an epic poem, relying on ChatGPT promotes the subtle implication that, in the end, my students’ own thinking and writing is of lesser value than the algorithmic output of AI. Reliance on AI will foster the idea that language—one of the greatest tools we have for connection, empathy, and union with others—is not really that important. What I hope to get across to my students, of course, is the opposite. I want them to know that their unique voice and way of thinking are important. It’s important that they write on their own not merely to pass a class or even develop better writing skills, but so that they develop an abiding sense that what they think is both important and worth sharing with others, even if it requires the difficult work of turning their thoughts into words. This is especially important for young people who are struggling to find their place in the world, anxious and insecure about what they have to give, and, ultimately, yearning for belonging and connection. Ultimately, what my students think matters; they need to know that it does and learn how to express it.

How might AI augment rather than detract from our capacity to create and share with others our works that connect us more fully to each other? I have no problem with ChatGPT taking over writing clichéd and superficial copy for soda companies and detergent brands, freeing us up to do more interesting work. But how do we ensure guardrails are in place so that it doesn’t take over the very things that make us distinctly human? Some see literal slavery as the gravest threat we face from AI, but—far more realistically and far sooner—we face the prospect of a servile dehumanization, of giving up activities that bring human beings a deep sense of joy, satisfaction, and connection.

When my students do write an essay or a short story on their own steam, after struggling through the research, crafting a cogent thesis statement, or finally sketching a character onto the page that only existed in their head before, many share with me the joy they feel in having done it, and done it on their own. And while it’s always easier to watch TikTok videos or Netflix series or, yes, ask an AI to do your homework for you, there is a special feeling of happiness that accompanies work that is hard, creative, and worthwhile for its own sake. And there’s a special pleasure, too, in being able to partake in such an experience as a grateful reader.