As Long As We’re All In Agreement

A writer or storyteller has an experience with the world, a relationship with oneself and others determined by circumstance, that informs or fuels a thought that provokes the next thought and so on. Ultimately, imagination is the summation of those reflections after many experiences and a relationship with the memory of those experiences. This ongoing thread of imaginative compulsions leads to the construction and eventual writing or telling of a story.

Moreover, a story is almost never fully formed when someone first begins to write or tell it. Its fullness comes from a storyteller’s ongoing relationship to language as it emerges from them, the possibilities of word choice and syntactical construction shaped by the subject at hand and, more largely, the physicality of the world in which they create, the setting of both the memory and of where the story is being written or told. The elements of one’s surroundings and hesitations in one’s breath are among many tangible realities that shape the writer or storyteller practicing their craft.

In other words, fiction, and storytelling more broadly, is a sum reflection of a human being’s physical experience (from reception to response, in motion or emotion) in their immediate world.

Erich Auerbach spoke to this line of reasoning directly in his seminal work Mimesis: The Representation of Reality in Western Literature. A thorough review of Western literary production, from Ancient Greece and Old Testament narratives to 19^th century literature, Auerbach argues that storytelling has been at its core a storyteller’s best attempt to imitate the reality of the physical world, through character, voice, interiority, causality, et al, appealing to a logic as much physical as it is emotional. This imitation is our means through which to understand our individual and collective experience of all that encompasses reality. It feels safe, therefore, to call these fictions and stories that emerge from the human imagination primary fictions, primary storytelling—primary in the way that historians usep to differentiate primary sources (e.g., diary entries, policy documents, personal notes, etc.) from secondary sources (i.e., the analysis or synthesis of primary sources). Primary suggests something more completely founded in firsthand historical experience, secondary as derivative of primary texts and further removed from direct historical experience.

This begs the question, of course, what then becomes a secondary fiction, a secondary story?

I want to pause here to clarify that these brief words are not an argument against accepting fictions and stories generated from large language models (and artificial intelligence technologies, broadly speaking) as pure fictions and stories. These brief remarks here are also not an argument in favor of receiving LLM/AI fictions and stories as pure to form as we’ve understood storytelling and creative work forms over the millennia. The fact that you must already be anticipating the distinctions I’ll be making between primary and secondary fictions and stories should have smartened you up to the latter.

What I do want to make certain before we go on is that all I’ll be asking of you after you read this is that you pause before wholly accepting digitally generated fiction and stories as a stand-in for what is otherwise creatively human. In other words, I’ve already resigned to the short- and long-term future in which the vast majority of us won’t have the tools or wherewithal to know how or even want to distinguish between primary and secondary fictions and stories. This feeling is only just an historical appreciation of the present. There are so many of us here, right now, who don’t have the tools or wherewithal to know how or even want to distinguish between credible sources of information and disreputable ones, between information determined by sequential or probabilistic reasoning and opinions masquerading as information discerned from concerted analyses.

I’m also not saying that we shouldn’t be distinguishing between the credible and disreputable, between information and opinion. To the contrary, and I’m sure you’ll agree, these distinctions matter deeply, as do the distinctions between primary and secondary creative works, and not simply because of one’s taste in stories, one’s preferences for voice or style or medium through which those stories are transmitted.

Benedict Anderson, in Imagined Communities: Reflections on the Origins and Spread of Nationalism, identified a handful of pillars essential to the formation of the nation-state and to the spread of nationalism and nationality as an essential component of one’s individual identity. He determined that there were but a few key elements that helped to coalesce otherwise distinct regions, dialects, languages, ethnicities, cultures, and traditions into these monoliths we call nations that we now take for granted. One of these most essential elements, he argues, is the novel, that malleable and even unwieldy and indeterminate version of the codex that, at heart, offers a hodgepodge of formal possibilities but which has largely been understood (and practiced) in the Western tradition as a realist, individualist narrative in which a protagonist (or set of protagonists) ambition to become agents of their own experience in the world. The novel, according to Anderson, became essential to national identity because of the social ideologies and priorities it could help to shape and spread, precisely because the fictions and stories contained within envisioned a larger community (and human experience) beyond the individual at the center of the book and which, most importantly to a reader’s experience, shared the ideologies and priorities of that individual across landscapes. In more plain terms, the novel demonstrated to its readership that the individuals in these stories (like the readers themselves) had the most important ethical, political, social, and cultural features in common with other individuals beyond their immediate experiences, imagining a wider and larger community that shared a singular destiny.

Coupled together, Anderson and Auerbach suggest why primary fiction, and primary storytelling and creative works more broadly, and our understanding of these works matters: all creatively human stories are firsthand imitations of a human being’s experience of embodied reality (causal, physical, emotional, conceptual, etc.) and that firsthand imitation has been a primary determinant of a nation’s and nationality’s ideological and physical borders. How we reflect our realities in creative work matters. Craft techniques (e.g., syntax, narrative elements, experimental shifts, various creative structures and descriptive texture, etc.) are used in service of that representation of reality. Form and content intertwine to weave the creative experience and the audience’s experience. Form cannot be extricated from content. We cannot compartmentalize a language and imagine that the partitioned manipulation of words and punctuation alone makes for a firsthand fiction or story or creative work.

Creative work generated by large language models will only ever be derivations of these primary works, defined here as a creative work generated by a human being’s sum experience with an embodied reality, even when that experience is limited to the physical constraints of their own bodies and the potential to lose that body. Creative work generated by algorithmic learning systems should only ever be understood as imitations of imitations, a text without a direct relationship to embodied reality, and even appreciated as a means through which to understand how human creativity has tended to use and construct language. These digital artifacts should never be received as a definitive practice of fiction and storytelling and creative work in themselves, never be received as a primary work.

We are, however, at a notable moment in our understanding of culture and creative practice. In the human story, we appear to be at the living cusp of the most naked of literary techniques. Personification, attaching a human quality to an inanimate thing—turning thick and roundly clouds into angry ones, for example; turning a flickering streetlight into a nervous one, for example—has transformed into an impulse to imagine that what these mechanical things we’ve made are making themselves look and sound and feel like the spitting image of things we have made and continue to make. We are starting to believe that the words and stories they string together look and sound and feel like the words and stories we string together. I choose to adapt the term “personification” in this moment because it feels approachable and actionable—and because I don’t like the word “anthropomorphizing”: it carries almost untouchable stakes, too high and perhaps because of those heights, the term itself has done little to persuade anyone to stop imbuing these large language models, these algorithms that continue to mutate beyond recognition and potentially beyond revision, with the best and worst qualities of our human selves.

In the human story, we’re at the cusp of believing that a digitally generated story is as fixed to our embodied reality as a human-crafted story. We are beginning to believe that the former isn’t just derivative, that by extracting form from content it isn’t just an imitation of our imitation of embodied reality. We’re at the cusp of failing to remember that large bodies of fictions and stories and creative work, produced and aggregated like this, aarrive to us with their own ideology of form and function and identity and community and time. That by mistaking secondary fictions and stories for primary ones, we are unintentionally beginning to shape our reality according to these subtly extreme, unrecognizable, and derivative reflections.

There are consequences to this, even if I can’t fully imagine a language to describe them. What I do feel capable of is the words for questions that can approximate the tangible impact of accepting and relying on digitally generated creative work and their reflections of our reflections of embodied reality.

For example, what will happen to us if (or when) the reading public (and larger audience for cultural work) no longer has the tools or wherewithal to know how or even want to distinguish a primary work from a secondary one? How will that gradually reshape what our conception of embodied reality is, and how will that in turn reshape what shared and collective experience is, what our imagined embodied communities are, what it means to be a nation, of nationality, of an ethnicity? How will that reshape what a shared history means, what a collaborative present becomes? How will our acceptance of (and even reliance upon) derivative language (our human languages reframed and reshaped through algorithms) gradually reduce the language at our sensory disposal, particularly because, in the end, computer algorithms always lean into efficiency and reduction as a product of probabilities that can be controlled? Will the imitation of the imitation of our embodied reality also reduce the complexity of human experience in the name of efficiency, we who rely on language to articulate and journey through who and what we are and become? Will algorithms author words out of our language that could otherwise heighten uncertain physical probabilities, words like “rebellion” and “doubt” and “black” and “queer”? With the algorithms determined by so few hands, developed and disseminated by so few relative to so many—in the absence of a democratic intervention or revision of these systems, who then becomes an authority on the subject of these systems, an authority on the subject of us and our languages that trained these systems? How does access to the backend of these systems grant greater and greater power to the financiers who developed them, and how does this narrow access grant wide latitude to disrupt, to reshape, the idea of community, of a nation, of the human story? How does this power threaten to reduce our access to and understanding of our primary creative works?

I don’t pose these questions because I imagine that answering them is possible or even reasonable. I ask these questions to attempt to reflect the unwieldiness of our present, of our current experiences with these technologies and with the uncertain anticipation of a future with these technologies so embedded in our lives. Questions, in the end, are the most valuable contribution to the human creative experience. Some questions take time to arrive at reasonable answers. Some questions just amount to a full stop after realizing that just understanding the questions renders us not yet fit for a response. This is good, I think. We can pause and take time to interrogate, to think. Questions drive our imaginations, but I don’t know that we can individually or collectively retain much of an imagination when we suggest that secondary creative works, imitations of imitations of our embodied reality, and the derivative textual and visual language they produce, can be stand-ins for primary creative works. I can only imagine what will happen to our perceptions of embodied reality, to this imagined community we call a nation, when we hand over more and more of our cognitive capacity to algorithms. What does it mean for the future, which should otherwise be built on individual human creative works meant to embody a collective human experience, when the fictions and stories and creative works we prioritize arrive via machines that can only offer a reductive imitation of our imitations?

My students know me to often quote dialogue from the Marvel Cinematic Universe as I teach, and there’s a line that Thor delivers after Thanos travels across universes for the final battle in Avengers: Endgame when the situation is tense and the ending is uncertain. Iron Man warns them that their predicament, Thanos waiting with his army at the ready, is just a trap. And, yet, Thor, exhausted and still grieving the loss of his people, sighs and says, “As long as we’re all in agreement.” The stakes in adopting these large language models and artificial systems across culture are so high, much or more than we can imagine capable of being lost in the process, not the least of which is the power and understanding of those embodied primary creative works that have come to determine who we are collectively and how we move forward together. My only hope is that as long as we’re all in agreement, as long as we recognize that not just the future that feels most precious and hopeful to us is possible but that all those futures in which many versions of bad or worse are also very likely, then let’s do what we do and continue attempting open discourse about these systems, about what their impact means and might mean for us today and for those of us tomorrow.