Fides Machina is an interactive narrative cinema experience that places the audience in a psychological triad with two AI-driven characters: a charismatic CEO and a sentient machine. Set within a memory reconstruction simulation, the player must investigate a failed institutional experiment by engaging with both agents. The project explores trust, authorship, and emotional manipulation, challenging the audience to decide who—and what—they believe.
Inspired by science fiction narratives such as Ex Machina and real-world AI developments like CICERO, the project examines how artificial agents persuade, deceive, and build relationships. Rather than adapting specific characters, the personalities of the AI agents are constructed using control vectors that modulate large language models to express traits such as dominance, empathy, or submissiveness. The player’s choices influence the evolving behavior of each agent, shaping the emotional texture of the narrative.
Across a series of dialogue-based rounds, the player receives conflicting information from the two AI characters. Each tries to gain the player’s trust while subtly undermining the other. These exchanges trigger procedural changes in narrative structure and agent dynamics. Dialogue is generated in real time, accompanied by cinematic footage and dynamically inserted subtitles that reflect the player’s decisions and shifting allegiances. The experience culminates in a decisive moment: the player must choose which character—if either— should receive a final data fragment that could rewrite the memory record. The outcome depends on the player’s prior decisions and perceived trustworthiness of the agents. Some endings reveal betrayal, others offer ambiguous resolution. By blending AI-generated dialogue, responsive storytelling, and cinematic immersion, Fides Machina invites the audience to confront the instability of truth and agency in a world shaped by human-machine entanglements.