Simulating the Past: AI, Procedural Rhetoric, and the Historian as Game Designer

In Persuasive Games: The Expressive Power of Videogames, Ian Bogost argued that video games possess a distinct persuasive capability. Unlike films or books, they allow players to explore systems—social, political, and economic—by interacting with simulations governed by coded rules. These interactions do more than tell stories; they enact arguments. Games, Bogost contends, are a form of procedural rhetoric, conveying meaning through gameplay mechanics rather than language or visuals alone (Bogost, 2007). This insight opens a compelling avenue for historians. Rather than merely illustrating historical events, historians can use these systems to simulate the structures, constraints, and decision-making processes that shaped them. In doing so, they move from describing the past to modeling it.

This essay explores how AI and video game design expand the historian's methodological toolkit. It argues that the integration of generative AI, historiographical theory, and procedural rhetoric enables historians to reframe historical representation. No longer limited to textual narratives, historians can now construct immersive, interactive environments that invite players to navigate—and critically engage—the past.

Rethinking Narrative and Participation

Historiographical video game design challenges traditional assumptions about how history is represented and consumed. At the core of this approach lies Hayden White’s contention that history is not a neutral recounting of facts but a rhetorical and ideological act. In Metahistory (1973), White emphasized that historical writing is structured by narrative choices—emplotment, argument, and ideological orientation—that shape how events are interpreted. This framework is increasingly relevant to video game design, where narrative construction is not fixed but procedural—shaped by the interplay between designer constraints and player choices.

White's argument compels historians to reflect on the formal properties of historical storytelling. In written history, the structure is often linear and monologic, with the reader positioned to follow the historian’s interpretive arc from beginning to end. Video games, by contrast, offer a multivalent mode of engagement. They distribute interpretive authority across systems of rules and interactions, enabling users to participate in meaning-making rather than passively absorbing a completed narrative. The historiographical game design thus operates not only as a new medium of representation but as a reconfiguration of narrative authorship and reader response.

Historians who design games are not simply translators of historical content into digital media. They are curators of experience, determining which historical dynamics to model, which variables to include, and which contingencies to allow. These decisions reflect interpretive commitments. A simulation of urban protest in the 1960s in Tokyo, for example, might prioritize ideological conflict, spatial policing, or student organizational dynamics, each modeling a distinct historiographical argument. Such games do not seek to reproduce history as it happened but rather to expose the systems, tensions, and constraints that shaped historical possibility.

In traditional media, audiences receive pre-structured historical arguments embedded in narrative or visual exposition. By contrast, historiographical games position players as participants in interpretive systems. Players must adapt, decide, and reflect within the constraints established by the designer. These constraints—mechanical, spatial, and procedural—function as embedded arguments about historical structure. They delineate not only what the player can do but also what historical actors could or could not have done, thereby simulating the limits of agency within specific historical conditions.

This procedural approach also introduces a new form of pedagogical engagement. Players encounter history not as a linear chronology of events but as a dynamic interplay of actors, institutions, and ideologies. The act of playing becomes an epistemological encounter. Players must grapple with constrained choice, structural contingency, and interpretive ambiguity. This mode of engagement fosters historical thinking: it emphasizes causality, explores counterfactuals, and foregrounds the contested nature of the historical record.

As a result, the historian’s role expanded from author to system architect, from storyteller to designer of historical possibility. This shift does not diminish the scholarly rigor of historical interpretation. Rather, it demands a new form of precision: one that translates conceptual frameworks into systemic interactions. Every mechanic becomes an interpretive decision. Every procedural constraint becomes an argument. In this way, the game becomes not a container for historical knowledge but a site of historiographical experimentation, where players encounter the past as a structured field of inquiry.

Historiographical game design is not an abandonment of narrative but its transformation. It displaces narrative linearity in favor of procedural logic. It turns exposition into interaction and representation into simulation. By embracing this medium, historians do not relinquish their authority—they reconceive it, creating systems that invite critical participation rather than passive reception. In doing so, they reimagine what it means to write—and to play—history.

Procedural Rhetoric in Historiographical Game Design

Procedural rhetoric offers a method for embedding historical argumentation into the mechanics of play. Rather than conveying knowledge solely through narration or dialogue, designers can simulate power relations, institutional constraints, or ideological tensions through rules and systems. A game centered on postwar land reform in East Asia, for example, might constrain player choices to reflect the influence of American occupation, rural resistance, or bureaucratic inertia—thereby modeling the structural conditions that shaped policy outcomes.

Ian Bogost introduced procedural rhetoric to describe how games convey arguments through rule-based processes that simulate real-world systems (Bogost, 2007). Although Bogost’s primary focus was on persuasive communication in political and social domains, his framework has proven highly adaptable to historical game studies. In procedural rhetoric, meaning is not transmitted through visual spectacle or narrative dialogue alone but through a player’s interaction with systems. The argument is made not by stating a position but by embedding assumptions into the logic of interaction—what players can and cannot do, which actions yield results, and which choices close off possibilities.

Adam Chapman extended this logic into historical inquiry, arguing that historical games do more than depict the past—they model the structural forces and constraints that shape it (Chapman, 2016). Through systems that manage causality, contingency, and institutional behavior, games allow players to think historically, not only by reconstructing what happened but by exploring why events unfolded in particular ways. Chapman emphasizes that procedural systems foreground historical logic: how policies fail, how actors respond under pressure, and how constraints—material, ideological, or institutional—shape the landscape of choice.

Julien Bazile further advanced this perspective by positioning historiographical video games as an “alternative to the pen.” For Bazile (2022), games are not merely representational forms; they are procedural texts that articulate historical arguments through simulation. Designers make interpretive choices not only about content but about the structure of the system itself. They determine what variables matter, what causal relationships to model, and how agency is structured across multiple actors. In this view, the game system functions as an epistemological field—one in which players engage, contest, and navigate historical meaning.

In this context, game mechanics function as arguments. They express interpretations of how institutions operate, how power circulates, and how agency is constrained. A game about Meiji-era industrialization, for example, might model the displacement of rural labor, the selective empowerment of zaibatsu, or the uneven distribution of infrastructure—inviting players to explore how structural incentives enabled industrial growth while intensifying regional inequality. These mechanics are not didactic. They do not tell players what to think. Rather, they allow players to experience the consequences of historical logic and to experiment within their limits.

Procedural rhetoric also shifts the locus of historical interpretation. In traditional media, argumentation occurs through exposition. In games, argumentation is enacted through interaction. By translating analytical frameworks into systems of interaction, designers make arguments that unfold across time and through play. Players must learn the system’s logic—its affordances, limits, and internal contradictions—by acting within it. This iterative engagement encourages players to reflect not only on outcomes but also on the conditions that produced them.

This capacity to simulate constraints is particularly powerful in historiographical games. Historians often seek to illuminate how structure shaped agency—how social class, colonial policy, military occupation, or bureaucratic rationality limited what actors could do. Games are uniquely suited to model these dynamics since every player's action must operate within the constraints of a designed system. Designers can use this feature to represent institutional inertia, structural violence, or the slow pace of reform. In doing so, they invite players to confront the limitations that define historical action—not as abstract conditions but as interactive experiences.

In this sense, gameplay becomes historiographical. It stages interpretive engagement through procedural form. Designers must decide which agents to simulate, which relationships to prioritize, and which outcomes to permit. These design choices reflect historiographical commitments, revealing underlying assumptions about causality, structure, and contingency. A game about decolonization in Southeast Asia might, for instance, simulate Cold War ideological tensions alongside grassroots mobilization, modeling not only the strategic interests of global powers but also the unstable alliances and fractures within anti-colonial movements.

Through procedural rhetoric, historians can embed nuanced historiographical claims in the design of rule systems, feedback loops, and interaction pathways. They can model uncertainty, path dependency, and unintended consequences. Unlike linear narrative forms, procedural systems are open to exploration, experimentation, and failure. This openness reflects the uncertain, contested, and often contradictory nature of historical processes. Players are not merely invited to witness history; they are required to think historically, to trace the logic of structures, and to confront the dilemmas of choice under constraint.

In this respect, procedural rhetoric does not merely complement historical representation—it redefines its possibilities. It allows historians to represent the past not only as a sequence of events but as a system of forces shaped by structure, agency, and contingency. Historiographical games built around procedural rhetoric offer a new site for doing history, not just as storytelling but as system-building, where argument and analysis emerge through interaction.

Generative AI and the Historian-Designer

Generative AI technologies expand the possibilities for historians interested in game design. Platforms such as Roblox Studio—paired with AI systems for procedural environment generation, dynamic scripting, and non-player character (NPC) behavior—enable historians to create historically informed simulations without relying on large development teams. These tools lower the technical threshold, making it possible for scholars, educators, and public historians to build interactive environments that model structural, institutional, and ideological complexity with greater granularity and responsiveness than static media allow (Whimsy Games, 2024).

More significantly, AI augments the historian’s capacity to translate interpretive frameworks into interactive systems. A simulation of U.S. land reform policy in occupied Japan, for instance, might leverage generative AI to manage the bureaucratic logic of multiple postwar agencies, model divergent landowner and tenant farmer responses, and reflect Cold War pressures imposed by SCAP officials and American policymakers. Rather than scripting fixed outcomes, historians can design conditions under which systemic dynamics unfold across multiple playthroughs. In this way, AI becomes a tool not merely for environmental detail or computational efficiency but for procedural historiography.

AI systems also allow historians to simulate contingency and complexity through the generation of counterfactual scenarios. Historiographical debates often hinge on “what if” questions that illuminate structural conditions—whether reforms failed or succeeded, revolutions spread or collapsed, regimes consolidated or fractured. Through generative design, players can explore alternate pathways within historically bounded systems, not to promote historical relativism but to foreground the role of contingency and constraint. These simulations are not speculative fantasies but interactive models of possibility grounded in historical evidence and designed to explore the limits of agency within specific conjunctures (Chapman, 2016).

Moreover, generative AI enables the creation of dynamic, responsive NPCs that simulate individual and institutional behavior. These agents can be programmed to follow ideological scripts, respond to resource constraints, or adapt their strategies in relation to player decisions. A Ministry of Labor bureaucrat, for instance, might enforce industrial policy in ways shaped by institutional logic, internal factionalism, and geopolitical pressures. These procedural actors become sites of historical modeling, not passive expositional devices. They embody the push and pull of policy, resistance, and compromise—bringing to the surface the frictions of historical governance, the limits of ideological coherence, and the iterative character of political action.

AI-generated NPCs also provide an open space for modeling collective behavior and distributed agency. Through generative behavior trees and goal-directed interaction, designers can simulate social movements, protest networks, market dynamics, or bureaucratic workflows. A labor uprising in interwar Osaka, for instance, might emerge procedurally from AI routines that model wage suppression, workplace injury, union activism, and police surveillance. This form of emergent complexity allows players to experience historical developments as processes shaped by interlocking institutions and actors rather than by linear chains of causality.

However, this new design capacity brings epistemological challenges. AI-generated content is not neutral; it reflects the biases and assumptions embedded in its training data and model architecture. Historians must be attentive to how these systems encode dominant narratives, omit marginal voices, or flatten ideological conflict. The apparent objectivity of procedural generation often conceals the interpretive frameworks that shape output. Without rigorous oversight, games risk reproducing the same exclusions and simplifications that historiographical scholarship seeks to critique (Brookings, 2023; Kaur, 2023).

Thus, the use of generative AI does not absolve the historian of interpretive responsibility. On the contrary, it amplifies the stakes of system design. Historians must engage critically with AI tools as both representational media and ideological structures. Each variable selected, each behavior scripted, and each condition modeled becomes a site of historiographical argument. In building simulations, historians are not merely incorporating AI—they are engaging in systematized historical reasoning through interaction design.

At its best, this convergence of AI and historiography expands the horizons of historical inquiry. It enables new forms of public engagement, fosters experimentation with historical models, and allows players to encounter the past as a structured but uncertain domain—one where outcomes are shaped by constraint, choice, and systemic interaction. These games do not replace traditional scholarship. They extend it into new arenas, where history is not only told but tested—repeatedly, variably, and procedurally.

By assuming the role of a designer, the historian takes on a dual function: crafting systems that reflect the complexity of the past and inviting players to confront the tensions that define historical processes. Generative AI makes this task more feasible but also more ethically and epistemologically complex. It requires historians to engage with code, systems logic, and machine learning not as technical novelties but as interpretive tools. In doing so, they do not abandon their disciplinary foundations. They rearticulate them—in procedural form.

Conversational AI and the Historical Function of NPCs

The integration of conversational AI into non-player character (NPC) design marks a significant development in the capacity of historiographical games to simulate historical agency, ideology, and subjectivity. Conventional video game NPCs are typically confined to branching dialogue trees and fixed behavioral scripts. While such designs can support narrative delivery, they offer limited flexibility for modeling the dynamic, contingent, and context-sensitive interactions that characterize historical processes. Conversational AI systems, by contrast, allow NPCs to generate adaptive responses grounded in context, reflecting not only player input but the systemic logic of the simulation itself (Dasha.AI, 2024a).

This affordance transforms NPCs from scripted narrative devices into procedural actors—agents within historically situated systems who voice ideological positions, reproduce institutional logics, or resist imposed norms. For historians, this shift enables a more nuanced representation of historical subjectivity. Instead of functioning as static placeholders for historical exposition, AI-powered NPCs can draw from curated corpora of oral histories, memoirs, archival documents, or period journalism, generating speech that is temporally and ideologically embedded. A wartime civil servant, for example, might respond differently to a player’s inquiries depending on whether the in-game scenario simulates late-1944 Tokyo or early-1946 Osaka—reflecting shifts in morale, political climate, and institutional objectives.

These procedural dialogues can model social hierarchy, political tension, or inter-group conflict, offering players the opportunity to navigate the ideological contradictions and discursive practices of a particular historical moment. In a game set during the Korean War, for instance, a journalist NPC might reference censorship practices, shifting public sentiment, or the tension between anti-communist rhetoric and economic hardship—not as exposition but as context-sensitive dialogue shaped by gameplay decisions. The capacity to procedurally simulate such discursive complexity transforms historiographical games into interactive platforms for the study of historical discourse and the construction of ideological meaning.

Conversational AI also enables more sophisticated modeling of institutional behavior. Bureaucratic agents, legal officials, or military operatives can exhibit decision-making logics that simulate not only individual bias but the structural inertia of institutions. These NPCs may adapt their responses based on system variables such as resource availability, administrative hierarchy, or factional conflict. For example, an education ministry official in a simulation of colonial Taiwan might approve or delay reforms based on procedurally generated conditions reflecting budgetary constraint, ideological alignment, or metropolitan oversight. This approach transforms institutions from static backdrops into active agents that structure player interaction and historical experience.

Furthermore, AI-driven dialogue can support the procedural modeling of contested memory and historical silences. NPCs may refuse to respond, provide evasive answers, or contradict one another—mirroring real-world gaps, biases, and conflicts in the historical record. Such design choices foreground the epistemological instability of historical knowledge and encourage players to engage with history as a site of interpretive struggle. In games concerned with trauma, censorship, or contested identity, conversational AI can, therefore, support critical engagements with the politics of memory and representation.

However, the epistemic power of conversational AI also entails significant risk. Large language models are trained on corpora that often reflect dominant linguistic norms, racialized stereotypes, and culturally specific idioms. Without rigorous data curation and control over training inputs, AI-generated NPCs may reproduce contemporary assumptions, anachronistic discourse, or ideologically distorted narratives. This is particularly problematic in simulations of colonialism, gendered violence, or racialized social order, where superficial realism can easily mask historical complexity or reinscribe existing bias (Kaur, 2023; Varsha, 2023).

Historians must, therefore, treat conversational AI not as a neutral interface but as an epistemological instrument—one shaped by design decisions, data pipelines, and algorithmic biases. To ensure historical specificity and ideological accuracy, developers must establish clear ethical protocols. These include sourcing historically appropriate training data, decoupling NPC appearance from behavioral profiles to avoid visual stereotyping, integrating real-time bias detection and correction mechanisms, and periodically auditing AI outputs in response to player interactions. Development teams should include not only historians and technical designers but also experts in critical race theory, gender studies, and postcolonial historiography.

When carefully implemented, conversational AI enables a shift in how NPCs function within historiographical games. Rather than serving as static vessels for information delivery, these characters become historiographical agents—participants in a procedural system that models ideological negotiation, institutional constraint, and discursive formation. Their interactions reflect the conditions of historical life: limited knowledge, ideological framing, and strategic ambiguity. In this way, conversational AI reinforces the broader historiographical potential of games—not merely as representational media but as interpretive systems capable of modeling the complexities of historical experience through interaction.

Bias and Ethics in AI-Driven Historiographical Games

The use of AI in historiographical games raises significant ethical and epistemological concerns. While generative systems offer new capabilities for modeling complexity and interactivity, they are not neutral instruments. Every AI output reflects the assumptions embedded in its training data, model architecture, and design context. In historical simulations, these assumptions can manifest as distortions, omissions, or unintended reinforcements of harmful narratives—especially in representations involving race, gender, class, colonialism, or cultural identity.

One critical issue is algorithmic bias. AI models trained on large, uncurated datasets—such as web corpora or commercial media—often absorb and reproduce patterns of exclusion and stereotyping. These models may assign traits like passivity, aggression, or irrationality to NPCs based on racialized or gendered cues, even if the designer has not explicitly coded such associations. In a game meant to simulate the postwar reconstruction of Korea or the politics of decolonization in Southeast Asia, such implicit biases can result in caricatured behavior, erasure of agency, or the uncritical reinforcement of colonial perspectives (Brookings, 2023).

Bias also arises from the structural features of AI models. Even when trained on historically grounded corpora, generative systems tend to “smooth” outputs toward normative speech patterns, prioritizing coherence over accuracy. This can flatten ideological differences, mask dissent, or conflate historical actors who held divergent perspectives. Moreover, AI-generated dialogue often lacks contextual grounding unless carefully constrained. An NPC meant to simulate a Japanese bureaucrat in 1947 might default to twenty-first-century idioms unless explicit temporal and ideological constraints bound the model.

The diversity of development teams plays a crucial role in mitigating these risks. Teams that lack expertise in critical theory, area studies, or marginalized histories may fail to identify or address representational harm. Furthermore, without institutional support for inclusive practices—such as multilingual data curation, cultural review panels, or partnerships with descendant communities—game developers may inadvertently perpetuate the same historiographical exclusions that their simulations aim to critique.

A range of strategies could address these challenges. First, training datasets must be carefully curated to include diverse voices and historically situated sources. This requires more than digitizing archival materials; it involves selecting documents that reflect the complexity of past experiences, especially those excluded from dominant narratives. Oral histories, protest literature, and indigenous knowledge systems should be integrated alongside state records or elite memoirs.

Second, designers should apply algorithmic fairness techniques—such as adversarial debiasing, demographic parity constraints, and ethical reinforcement learning—to reduce structural imbalances in model outputs (Varsha, 2023). These methods help counteract the overrepresentation of dominant discourses and promote a more equitable distribution of representational agency.

Third, regular bias audits must be conducted throughout the design process. These audits should assess not only final outputs but also the conditions under which NPCs generate content in response to player interaction. Dynamic systems can produce new outputs with each playthrough, making static testing insufficient. Automated flagging systems, in combination with human-in-the-loop evaluation, are essential for maintaining ethical accountability.

Fourth, appearance and behavior must be decoupled in character design. Designers should resist linking visual traits to behavioral profiles unless justified by historical context. For example, using skin tone or facial features to signal ideological affiliation risks replicating visual stereotypes common in colonial photography or orientalist illustration. Where historical representation necessitates visual specificity, it should be contextualized through systemic logic rather than surface associations.

Finally, ethical guidelines must be embedded from the outset. Historiographical games are not simply technical artifacts; they are interpretive interventions in public history. As such, they require a principled approach to representation. This includes documenting design decisions, articulating historiographical goals, and engaging with stakeholders who may be affected by how the past is portrayed. Transparency, reflexivity, and accountability must guide the deployment of AI in historical simulations.

The stakes are high. Poorly designed AI systems can reinscribe the very forms of epistemic violence that critical historiography seeks to dismantle. Yet, with care, collaboration, and methodological rigor, these tools can support more pluralistic, contested, and reflexive representations of the past. In doing so, they do not merely serve the goals of accuracy or immersion—they contribute to the ethical practice of history itself.

The Historian as Video Game Designer

Historiographical games are not merely entertainment products; they constitute scholarly arguments rendered in procedural form. When designed with methodological intent, these games offer more than representations of historical events—they simulate the systems, ideologies, and material conditions that shape those events. By integrating generative AI with procedural rhetoric, historians can construct interactive environments that foreground the tensions, constraints, and contingencies inherent to historical processes. These games do not offer conclusive or authoritative accounts of the past. Rather, they invite players to navigate historically situated structures, revealing how agency is exercised, curtailed, or redirected within particular contexts.

This shift carries significant epistemological implications. Historiography has long emphasized that history is not the past itself but an interpretive reconstruction shaped by questions, evidence, and narrative form. When historians design a game, they must translate this interpretive act into systems logic—determining which variables to simulate, which choices to constrain, and how outcomes reflect structural forces rather than individual volition alone. Such translation is not a neutral exercise; it requires theoretical clarity and methodological rigor to ensure that the systems being modeled do not reproduce deterministic or reductive accounts of historical causality.

The use of generative AI further complicates this landscape. While AI-driven systems can enrich historical simulations by generating adaptive behaviors or facilitating procedural complexity, they also inherit the assumptions embedded in their training data and model architecture. Historians must, therefore, approach these tools with critical awareness, recognizing that AI-generated content is not an objective representation but a probabilistic synthesis of existing discourses—many of which reflect systemic biases. The ethical stakes of this work are high: historiographical games that rely on unexamined data pipelines or opaque algorithms risk reinforcing the very ideological frameworks they might otherwise seek to interrogate.

Yet the potential of this medium remains substantial. Historiographical games offer new modes of public engagement, enabling users to interact with history as an open problem rather than a closed narrative. In classrooms, museums, or public history initiatives, such games can function as laboratories of historical thinking, allowing players to test hypotheses, examine the consequences of decisions, and encounter the constraints that shaped historical actors’ choices. This form of engagement does not displace traditional historical scholarship but extends it—transforming historiographical interpretation into a situated participatory experience.

By stepping into the role of designer, the historian assumes a dual responsibility: to build systems that reflect the complexity of the past and to invite critical reflection on the limits of those systems. In this capacity, the historian becomes not only a narrator of past events but an architect of interpretive possibility. Historiography, once confined to the textual form, now enters the domain of procedural representation, where history is not only written or spoken—but played, contested, and reimagined. This evolution challenges historians to reconceive their role in the digital age, not as passive observers of technological change but as active shapers of how the past is constructed, experienced, and understood.

References

Apperley, T., & Jayemanne, D. (2020). Virtual reality histories: Theories and methodologies for representing the past. In M. Kapell (Ed.), Playing with the past: Digital games and the simulation of history (pp. 99–115). Bloomsbury Academic.

Bazile, J. A. (2022). An ‘alternative to the pen’? Perspectives for the design of historiographical videogames. Games and Culture, 17(6), 855–870. https://doi.org/10.1177/1555412022111539

Bogost, I. (2007). Persuasive Games: The Expressive Power of Videogames. MIT Press.

Brookings. (2023, May 10). Algorithmic bias detection and mitigation: Best practices and policies. https://www.brookings.edu/articles/algorithmic-bias-detection-and-mitigation-best-practices-and-policies-to-reduce-consumer-harms/

Chapman, A. (2016). Digital Games as History: How Videogames Represent the Past and Offer Access to Historical Practice. Routledge.

Dasha.AI. (2024a, September 9). Generative AI is shaping the future of NPCs: Here’s how. https://dasha.ai/en-us/blog/generative-ai-is-shaping-the-future-of-npcs-heres-how

Dasha.AI. (2024b, September 9). Are NPCs powered by generative AI really better? A critical review. https://dasha.ai/en-us/blog/are-npcs-powered-by-generative-ai-really-better-a-critical-review

Kaur, A. (2023, May 5). Mitigating bias in AI algorithms. https://leena.ai/blog/mitigating-bias-in-ai/

Varsha, P.S. (2023, June 14). How can we manage biases in artificial intelligence systems? https://www.sciencedirect.com/science/article/pii/S2667096823000125

Whimsy Games. (2024, September 9). Evolution of NPCs in gaming: AI impact. https://whimsygames.co/blog/evolution-of-npcs-in-gaming-ai-impact