���ϳԹ���Դ

In a world increasingly formed by generative artificial intelligence (AI), virtual reality (VR), and interactive media, the research project Extending Digital Narrative (XDN) at the Center for Digital Narrative engages with an ever-changing environment. This team of artist-researchers explores how creative experiments in emerging technologies can generate new insights into what it means to tell stories in the digital age.��

Those in the XDN project immerse themselves in the tools of the future to understand how the narratives are shaped, undertaking artistic practice as a method of inquiry. Through five case studies, the team shows how hands-on experimentation reveal the hidden assumptions, biases, and affordances in these new technologies in a recent article.

Reanimating the Past: Lina Harder’s Hedy Lamarr Chatbot��

Lina Ruth Harder’s project is a generative AI chatbot designed to simulate Austrian-American actress and inventor Hedy Lamarr. Visitors to the “More Than Meets AI” exhibition in Bergen could write with “Hedy”, who presented herself in a 1950s-style virtual café. Users engage in short text conversations in English or German, discussing both Lamarr’s life in Hollywood and her work on radio guidance systems for torpedos.��

“I call these histobots: conversational agents driven by generative AI that re-enact persons of historical record for edutainment purposes,” Lina writes of her project. Harder’s chatbot was more than a novelty act, it was a critical probe into how AI systems reconstruct historical figures. Even drawing on 64 carefully crafted prompts to encode Lamarr’s voice and personality, the results were still riddled with “GPTisms” – hallucinated facts, anachronisms, and overly polished language. Despite being set in 1951, the bot would sometimes refer to smartphones.

The Lamarr chatbot became what philosopher Jean Baudrillard would refer to as a simulacrum, a copy without an original. The work highlighted how AI can distort historical memory, flatten complex lives into digestible, sanitized narratives, and reinforce existing cultural biases – especially around gender and race.��

Harder counters the marketing for apps for similar historical chatbots, these are not “encounters with the past”: “At best, they serve as probes. They reveal how storytelling bends when authored by algorithms, how platforms curate cultural memory, and how histories of yesterday and fantasies of tomorrow become entangled in infrastructures beyond our control.”

Harder’s work offers additional insight into a key finding of the XDN team – that building with AI is not a neutral act. For her, the point is not that the chatbot failed, but what its failures reveal. The project exposed how generative AI systems are shaped by training data, platform constraints, and the biases embedded in both.

Satire in the Latent Space: Scott Rettberg’s “Republicans in Love”��

Scott Rettberg’s “Republicans in Love” is a political satire told through text-to-image and text-to-video AI generation. Using platforms like OpenAI’s DALL·E and Runway, Rettberg created hundreds of image-text pairs that show the absurdities of contemporary American politics, particularly the rise of Trumpist populism. Images are generated from a single prompt, surreal or ironic sentences like “Republicans in love, angry about the news, eating greasy cheeseburgers at the President’s desk in the Oval Office, in the style of Caravaggio.” The results are strange, evocative, and often hilarious. In Rettberg’s view they also serve as a form of “cyborg authorship”:

“That is, the idea that storytelling with generative AI is ontologically different from traditional forms of writing, because both human intelligence and advanced computational processes contribute to the creation of the narrative experience.”

He suggests text-to-image generation is not just a tool, but a new writing environment. The prompt becomes a kind of literary form – part aphorism, part performance. The AI doesn’t just illustrate the text; it transforms it, adding layers of meaning and unpredictability.

As the technology evolves, so do the constraints. Later versions of DALL·E introduced content moderation and prompt rewriting, making it harder to produce politically charged or stylistically experimental images. Rettberg notes a shift toward photorealism and away from the surreal, glitchy aesthetics that once made AI art feel genuinely new. In addition to the satirical elements, he sees the work as “rapid media archaeology” capturing fleeting moments in the evolution of AI aesthetics – moments that may be lost to corporate sanitization.��

Despite these challenges, Rettberg sees potential in AI satire as a form of resistance. While right-wing groups also use generative tools to spread propaganda, projects like “Republicans in Love” do offer a counter-narrative:��

“If nothing else, I find the process of making these images cathartic, and a means to cope.”

One Artist, Many Voices: David Jhave Johnston’s “Messages to Humanity”��

What happens when a single person can make a film with no crew, no actors, and no studio? David Jhave Johnston’s “Messages to Humanity” explores the question through a series of 30 short videos created using Runway’s Act-One, a generative AI tool that maps a performer’s voice and facial expressions onto animated characters.

Each video is a poetic, philosophical reflection on the state of the world – climate collapse, war, AI acceleration – delivered by a rotating cast of surreal avatars. Some are human, others alien or more abstract.

The videos are short, often under a minute, but they do not hold back. One message declares: “All narratives end. Bodies die. Civilizations collapse.” Johnston’s project is both a technical experiment and a personal way to navigate a perilous present, while suggesting alternatives. It shows how AI can democratize filmmaking, allowing solo creators to produce larger narratives with several character.

But it also raises questions about authorship, identity, and the role of the artist in a world where anyone can become anyone. Are new forms of critical engagement needed now? What does it mean to speak to humanity through a synthetic face? What responsibilities come with that power?��

Act-One, Johnston argues, represents a rupture in the history of cinema. What was once a time-consuming process of cameras, scheduling, actors, direction and time – generative AI replicates a significant subset of the endeavour through one person. The technology does not erase barriers, but it lowers them dramatically. But is it as good as the studios?��

Johnston answers: “Yes and no. For those on a limited budget, it is impeccable; for those with a studio budget, it is of questionable quality.”

Empathy Through Immersion: Sérgio Galvão Roxo’s “Her Name Was Gisberta”��

In “Her Name Was Gisberta,” Sérgio Galvão Roxo uses VR to tell the story of Gisberta Salce, a Brazilian trans woman murdered in Portugal in 2006. The project is a powerful example of how immersive media can foster empathy and social change.��

The VR documentary places viewers inside Gisberta’s world, from her childhood to her tragic death. Using 360-degree animation and narration by a Brazilian trans voice actress, the experience invites users to “storylive” her life. The goal is not to shock, but to humanize and create a space for reflection, understanding, and emotional connection.��

Roxo draws on research in social psychology and human-computer interaction to design the experience. He employs “virtual reality perspective-taking” techniques, allowing users to see the world from Gisberta’s point of view.

“The feedback was seen as cathartic for some viewers, as the frequency of viewers crying indicated a form of emotional release from the weight of the project, confirming its potential to breach barriers,” writes Roxo.

“By addressing empathy from another perspective, we can see it not as an innate or predetermined quality, but as a “muscle”."

The project is also a memorial. It honours not just Gisberta, but over 1600 victims of transphobic violence in Portugal and Brazil.��

Roxo’s next project will build on this work, using VR to document and raise awareness about the practices commonly known as conversion therapy in Europe, with testimonies from survivors. The project was launched summer of 2025, as the website .

Rethinking LLMs in Games: Haoyuan Tang’s Narrative Mechanics��

In most modern games, players interact through actions – moving, fighting, exploring – not through dialogue. Writing natural language as a player is quite rare.��

While many game developers explore how to add chatbots into non-player characters (NPCs), Haoyuan Tang takes a different approach: what if AI in games didn’t listen to what players say, but to what they do?

“Games are very good at providing structure without language,” writes Tang. “LLMs, by contrast, are very good at producing language that doesn’t necessarily require a specific structure.”

Tang is developing a prototype in Unreal Engine 5, which logs the player’s actions translating them into narrative prompts for a large language model. The goal for the AI is to generate story content – dialogue, environmental descriptions, character reactions – to fit the player behaviour.��

“In this view, AI is treated as a narrative agent,” writes Tang, embedding AI within the game’s existing logic, rather than forcing it to speak a language the game doesn’t understand.

The approach depends not just on technical sophistication, she writes, but on careful design. The AI must be tuned to the genre, platform, and player expectations. A poetic NPC might work in a fantasy RPG but feel jarring in a horror game.

Why Practice-Based Research Matters��

If we want to understand how AI and immersive media are changing the way we tell stories – and the way we understand ourselves – we need get our hands dirty.��

The XDN team’s work challenges traditional boundaries between art and scholarship. In many academic settings, creative practice is still seen as separate from “real” research. But as the authors argue, when it comes to studying emerging technologies, theory alone isn’t enough: “Going under the hood, and actually creating creative work in these environments, offers a deeper understanding of how the technologies themselves are evolving, and allows for more insight in their constraints and affordances for narrative.”

By building, testing, and performing with AI and XR tools, these researchers gain insights that would be impossible to obtain from the outside. They show that storytelling is not just about content, but about platforms, interfaces, and infrastructures. And they demonstrate that the future of narrative will be shaped not just by writers and artists, but by engineers, designers, and algorithms:��

“We advocate for bridges between theory and practice, particularly as new media, games, AI, and XR continue to reshape storytelling.”

About the Authors��

• Lina Ruth Harder is a PhD Fellow at the Center for Digital Narrative, University of Bergen, exploring AI-driven historical personas and the ethics of digital memory.
• David Jhave Johnston is a postdoctoral researcher and digital poet at the Center for Digital Narrative, known for his experimental work in AI-generated digital poetry
• Scott Rettberg is the Director of the Center for Digital Narrative and a pioneer in electronic literature
• Sérgio Galvão Roxo is a PhD Fellow at the Center for Digital Narrative and film maker, focusing on immersive media and VR as tools for social education and activism.
• Haoyuan Tang is a PhD Fellow at the Center for Digital Narrative, researching the integration of generative AI into video game narratives through player interaction��

Reference

Harder, Lina Ruth, David Jhave Johnston, Scott Rettberg, Sérgio Galvão Roxo, and Haoyuan Tang. 2026. "Extending Digital Narrative with AI, Games, Chatbots, and XR: How Experimental Creative Practice Yields Research Insights" Humanities 15, no. 1: 17. ��