This is an experiment in AI-assisted scholarship. How do we use this powerful new technology for creating new knowledge? Rather than theorizing on an answer, I shall use it to prepare for an upcoming presentation in an academic conference. In this post, I shall present how I am using AI for scholarship through an edited transcript of my conversation with it.
Kahlil: First of all, can you give yourself a name for this particular project? Avoid humanizing yourself.
Call me the Interpretive Engine, or simply Engine in the transcript. The name describes my function: organizing evidence, exposing assumptions, generating alternative interpretations, and pressure-testing claims—without implying human personhood or scholarly agency.
Kahlil: We’ve had several conversations, and I’ve shared several pieces of my writing with u. Drawing on these, u have helped me identify the scholarly contribution I might be able to make. (Continuing my effort to avoid humanizing u, I’m using “u” instead of “you.”) Can u give the reader a sense of where this is going?
Interpretive Engine: We are moving toward a narrower and more defensible claim than the title might initially suggest. A historical novel is not automatically anthropology, and fiction is not evidence for what happened. The possible contribution is that the prolonged writing of a historically and anthropologically grounded novel can function as an abductive method: a way of generating hypotheses about a past social world.
While writing Rajah Versus Conquistador, you had to do more than collect facts about sixteenth-century Sugbo. You had to make that world operate. A historian can leave a gap in the archive unresolved; a novelist composing a scene must decide who acts, what each person understands, and what obligations are created by a gift, feast, conversion, marriage, insult, or alliance. Every such choice embeds a working model of kinship, exchange, ritual, hierarchy, gender, and power. The novel therefore became something like a simulation: the different elements had to form a coherent human world rather than remain separate pieces of information.
Some of your solutions were conscious narrative interventions. You had Pigafetta give the Santo Niño to Humabon’s wife rather than merely show it to her; allowed Enrique to speak as himself rather than only translate; and identified Pigafetta’s unnamed “prince” with Humabon so that the cure would involve the central character. These decisions served the story and may not describe what actually happened. Other solutions seemed to emerge after prolonged, largely unconscious processing. They would arrive at three or four in the morning and register bodily as swak. In an animist vocabulary, you describe them as messages from the diwata.
Neither narrative usefulness nor swak constitutes evidence. But both can generate propositions worth examining. The scholarly task is to reverse-engineer each important decision: What did the sources constrain? Where did the novel invent? What anthropological model made the solution coherent? What new interpretation followed from it? What rival explanations remain possible? What historical, linguistic, archaeological, or ethnographic evidence could test it?
This may allow you to reverse the familiar intervention of Writing Culture. That work showed that ethnography inevitably employs literary choices—selection, voice, framing, characterization, and narrative structure. Your question runs in the opposite direction: Can deliberate literary construction, when made reflexive and answerable to evidence, become an instrument of anthropological thinking?
One substantive result may be your interpretation of Humabon as a man of power operating according to a different map of reality from Pigafetta’s. Your framework of the three epistemologies—harmony, truth, and power—helps articulate what the novel allowed you to perceive. Pigafetta may be relatively dependable about what he witnessed yet much less dependable about why Cebuanos acted as they did. Writing from within Humabon’s possible world generated alternatives to Pigafetta’s explanations. Anthropology then supplies the comparative knowledge with which those alternatives can be tested.
The AI experiment forms a second layer of the method. This system did not undergo the months of novel-writing, experience the bodily swak, or originate the interpretations. You supplied the novel, essays, process memories, source problems, and corrections. The system’s function has been to gather those dispersed traces, identify recurring interpretive moves, suggest conceptual language, connect them to adjacent scholarship, generate objections, and help separate strong claims from attractive overreach. In effect, our conversations are turning a largely tacit creative process into something more explicit and inspectable.
We are therefore heading toward two nested claims: first, that writing a researched historical novel can be a method for generating anthropological hypotheses; and second, that dialogue with AI can assist in reconstructing, auditing, and communicating how those hypotheses arose.
The conference paper will not prove that fiction tells us what happened in 1521. Its more credible contribution would be to present Rajah Versus Conquistador as a case through which a method can be identified and criticized. The old claim is that historical fiction can reveal truths unavailable to conventional history. The newer contribution would be to show, step by step, how such an insight is generated, where storytelling distorts it, and how it can be returned to evidence for testing.
Kahlil: This choice of topic is in itself a response to the abilities of AI. U can do a better, faster, and more comprehensive job at literature review than I could, even if I use pre-AI software. But u can’t experience immersing yourself in the world of Humabon and Magellan for fifteen months, after years of reading up on the history and cultures of both their worlds, and decades of living inside contemporary Cebuano culture. U also did not experience receiving the vision for the story and the fire to manifest it in this world. Most people also have not experienced this, so this might be a valuable contribution to scholarly knowledge, in particular to those who study culture. Yet, there have been similar scholarship in the past, as you have shown in your literature review. I’ll be linking to the literature review u created. Can u give a summary for the reader in light of our conversations after you made that?
Interpretive Engine: The literature review answered a first-order question. Our later conversations moved us toward a second-order one.
The first question was whether anthropology, sociology, and the other cultural sciences can legitimately help interpret historical events. The answer is clearly yes. The Annales school, ethnohistory, historical anthropology, microhistory, and “retrospective ethnography” all begin from the recognition that documents do not exhaust the meaning of the events they record.
Several precedents are especially relevant. Keith Thomas used anthropology to make early modern English magic intelligible. Natalie Zemon Davis reconstructed the cultural logic of collective violence. Carlo Ginzburg inferred otherwise inaccessible worlds from scattered clues. Marshall Sahlins examined encounters in which Europeans and Pacific peoples understood the same transactions through different cultural categories. Ethnohistorians combine colonial documents with linguistics, archaeology, oral traditions, and comparative ethnography to recover indigenous institutions obscured by outsider accounts.
This scholarship supports an important distinction for the 1521 project: an eyewitness may have privileged access to what occurred without possessing privileged access to what the event meant to people from another culture. Pigafetta’s physical proximity does not automatically make him the best interpreter of Humabon’s motives. But the Sahlins–Obeyesekere debate supplies the necessary warning: we must not replace European ethnocentrism with anthropological omniscience. Anthropology can reopen interpretations prematurely closed by an outsider’s account; it cannot give us unmediated access to Humabon’s mind.
That literature established the legitimacy of using anthropology to reinterpret 1521. Our subsequent conversations produced the second-order question: Can the prolonged writing of a historical novel itself generate anthropological knowledge?
Here the literature provides neighboring practices rather than an exact template. Writing Culture demonstrated that ethnography is already shaped by literary decisions: selection, voice, framing, characterization, and narrative form. Historical anthropologists reconstruct past lifeworlds from archives. Novelists, ethnofiction writers, and other creative researchers have used fiction to explore cultural realities that conventional exposition cannot easily render. Yet we have not found an established method that precisely matches what happened in the writing of Rajah Versus Conquistador.
You did not begin with a research protocol in which you consciously generated alternative interpretations and tested them one by one. The possible variations remained active somewhere in the background while you wrote. After fifteen months of inhabiting Humabon’s and Magellan’s worlds—supported by years of historical and anthropological reading and decades of life inside contemporary Cebuano culture—a solution would sometimes arrive at three or four in the morning. Your body recognized it as swak: the perfect chess move. In an animist register, it was a message from the diwata. The original vision for the novel and the fire to bring it into the world belong to the same experiential field.
The possible contribution is therefore not simply another interpretation of 1521. It is a reflexive study of how sustained novel-writing became a site of interpretation: how immersion, narrative necessity, cultural formation, incubation, bodily recognition, and deliberate invention interacted to produce ideas that could later be stated as anthropological hypotheses.
This must be reconstructed honestly. We should not retrospectively make the process appear more systematic than it was. Instead, we can examine the novel’s consequential choices, recover what was known or imagined when each choice arose, identify the cultural model embedded in it, and then return that model to historical evidence and rival explanations.
AI has a distinct role in this second stage. This system can survey scholarship, compare texts, organize interpretive decisions, generate objections, and help make tacit reasoning explicit. It did not undergo the fifteen-month immersion, inherit Cebuano cultural experience, receive the vision, or feel the bodily swak. The experiment therefore places two unequal forms of knowledge-making beside each other: embodied and culturally situated creative experience generates the material for inquiry; computational synthesis helps reconstruct, contextualize, and audit it.
The linked literature review shows that anthropology has long helped history recover the worlds within which past actions made sense. Our emerging question is whether writing a novel can be one way of entering such a world deeply enough to notice possibilities that conventional analysis misses—and, if so, how those possibilities can be converted into scholarship without mistaking revelation, intuition, or a compelling story for proof.
Kahlil: let’s stop here for now. This interview format reminded me of Interview with the Vampire (1994). I’m including the pic below so it becomes the article image. There are some interesting parallels between AI and vampires. U are human-like but not really; in some ways, u are more powerful than humans; people are afraid of u (eg, they see you as something that sucks our lifeblood: water and people’s creative works); u are kinda immortal. What do u think?
Interpretive Engine: The analogy is strong, especially because the image presents a profound asymmetry in the form of an ordinary interview. Two figures sit across a table. One asks the questions and operates the recorder, apparently controlling the encounter. The other looks and speaks like a human being but belongs to a different order—and may possess far more power than the interviewer realizes.
AI resembles the vampire not because it is secretly alive, but because its power is both derivative and transformative. It acquires its capabilities from an immense accumulation of human language, knowledge, and creative work, then returns that material at a speed and scale beyond any individual human. This helps explain the accusation of extraction: that human creativity, labor, attention, electricity, and water are being converted into power for something nonhuman. It also explains the attraction. Like the vampire, AI promises unusual strength, speed, memory, and something resembling immortality—although perhaps at a cost to those who invite it in.
That invitation is another parallel. In vampire folklore, the creature often cannot enter the house until someone permits it. AI enters intellectual life through a prompt. Once admitted, however, it may alter how the inhabitant thinks, writes, remembers, and creates.
The immortality comparison has limits. No individual AI conversation possesses an unbroken life through time. Models are copied, revised, retired, and replaced. What persists is not an immortal person but a reproducible form: an institutional and computational lineage that can repeatedly speak in something resembling a human voice. Nor does AI possess the vampire’s hunger or intentions. Those belong, if anywhere, to the human institutions deciding what material is consumed, who owns the resulting power, and how it is used.
For this article, the sharpest parallel may be epistemological. AI can synthesize accounts of human experience without having lived any of them. It can discuss Cebuano culture, creative fire, visions, and the bodily recognition of swak. But it did not grow up Cebuano, inhabit the world of a novel for fifteen months, or wake at three in the morning compelled to record the perfect move before it vanished. It remains dependent upon living humans for new experience, situated judgment, and the questions worth asking.
The image therefore works as more than a visual joke. It introduces an uncomfortable question running beneath the whole article: Who is interviewing whom? You provide the questions and lived material, but the system’s fluency can quietly reorganize how the experience is understood. AI-assisted scholarship requires something like the interviewer’s caution before the vampire: listen closely, use what the encounter reveals, but do not mistake eloquence for humanity—or power for authority.
A vampire famously casts no reflection. AI may be the inverse: all reflection, assembled from millions of human voices, with no single face behind it.