The Machine in the Mirror: SEED CCO on AI’s Role in Digital Autocracies

At the center of Klang Games’ ambitious society-sim SEED lies a provocative experiment: what happens when you give players total agency over a persistent, AI-driven world? Speaking at a recent preview, DualShockers sat down with CCO Ívar Emilsson to dissect the game’s controversial reliance on generative AI and a recurring, somewhat unsettling player behavior—the drive to build rigid dictatorships. While SEED aims to simulate life on the planet Avesta with tens of thousands of autonomous "Seedlings," the emergence of authoritarian regimes has become a focal point of the studio’s internal research into human-AI dynamics.

Emilsson defended the integration of generative AI not as a cost-cutting measure, but as a bridge for emotional resonance. He noted that players can communicate directly with their Seedlings, who possess distinct needs, relationships, and daily routines that persist even when the user is logged out. This persistent autonomy, powered by the technical foundations of Electronic Arts’ SEED research group, allows characters to "live" 24/7. However, this same depth has led to a fascination with power concentration. According to Emilsson, the popularity of dictatorships among the fanbase often stems from a desire for efficiency and order in a complex, high-stakes simulation where democratic consensus can feel messy and slow.

The Allure of the Digital Iron Fist

Behind the Scenes: The phenomenon of players gravitating toward autocracy isn't just a quirk of the SEED community; it’s a reflection of how gamers prioritize control in environments where they feel a heavy "emotional attachment" to their virtual charges. Emilsson pointed out that because the stakes in SEED involve the survival and happiness of specific characters, players often view democracy as a liability. In a world where AI agents have their own desires, a central authority provides a shortcut to progress that many find irresistible, even if it mirrors the very systems they might critcize in the physical world.

This trend has forced the developers to look closely at the ethical guardrails of their tech. As industry leaders like those at GamesBeat have noted, Klang is attempting an "ethical approach" to AI, ensuring that these digital lives aren't just empty shells. By giving characters the ability to remember past interactions and form independent opinions, the simulation becomes a mirror. When a player chooses to rule as a dictator, the AI reflects the social friction and the eventual erosion of the seedling's individual agency, turning the game into a sandbox for exploring the consequences of power.

The technical pedigree behind SEED—led by veterans from CCP Games—suggests that the team is comfortable with "emergent" gameplay that isn't always pretty. Just as EVE Online became famous for its corporate espionage and wars, SEED is poised to become a case study in how AI can amplify social engineering. Emilsson's comments highlight a shift in game design: the goal is no longer just to give the player a power fantasy, but to use generative technology to make that power feel heavy, complicated, and perhaps a little too real for comfort.

At the center of Klang Games’ ambitious society-sim SEED lies a provocative experiment: what happens when you give players total agency over a persistent, AI-driven world? Speaking at a recent preview, DualShockers sat down with CCO Ívar Emilsson to dissect the game’s controversial reliance on generative AI and a recurring, somewhat unsettling player behavior—the drive to build rigid dictatorships. While SEED aims to simulate life on the planet Avesta with tens of thousands of autonomous "Seedlings," the emergence of authoritarian regimes has become a focal point of the studio’s internal research into human-AI dynamics.

Emilsson defended the integration of generative AI not as a cost-cutting measure, but as a bridge for emotional resonance. He noted that players can communicate directly with their Seedlings, who possess distinct needs, relationships, and daily routines that persist even when the user is logged out. This persistent autonomy, powered by the technical foundations of Electronic Arts’ SEED research group, allows characters to "live" 24/7. However, this same depth has led to a fascination with power concentration. According to Emilsson, the popularity of dictatorships among the fanbase often stems from a desire for efficiency and order in a complex, high-stakes simulation where democratic consensus can feel messy and slow.

The Allure of the Digital Iron Fist

Behind the Scenes: The phenomenon of players gravitating toward autocracy isn't just a quirk of the SEED community; it’s a reflection of how gamers prioritize control in environments where they feel a heavy "emotional attachment" to their virtual charges. Emilsson pointed out that because the stakes in SEED involve the survival and happiness of specific characters, players often view democracy as a liability. In a world where AI agents have their own desires, a central authority provides a shortcut to progress that many find irresistible, even if it mirrors the very systems they might criticize in the physical world.

This trend has forced the developers to look closely at the ethical guardrails of their tech. As industry leaders like those at GamesBeat have noted, Klang is attempting an "ethical approach" to AI, ensuring that these digital lives aren't just empty shells. By giving characters the ability to remember past interactions and form independent opinions, the simulation becomes a mirror. When a player chooses to rule as a dictator, the AI reflects the social friction and the eventual erosion of the seedling's individual agency, turning the game into a sandbox for exploring the consequences of power.

The technical pedigree behind SEED—led by veterans from CCP Games—suggests that the team is comfortable with "emergent" gameplay that isn't always pretty. Just as EVE Online became famous for its corporate espionage and wars, SEED is poised to become a case study in how AI can amplify social engineering. Emilsson's comments highlight a shift in game design: the goal is no longer just to give the player a power fantasy, but to use generative technology to make that power feel heavy, complicated, and perhaps a little too real for comfort.

The Algorithmic Compliance Trap

Reading Between the Lines: The developer's insistence that AI creates emotional resonance is a fascinating claim, but it glosses over the inherent contradiction of generative systems: they are designed to please. If a Seedling’s Large Language Model (LLM) is optimized to respond in a way that feels "correct" to the player’s input, it may inadvertently reward authoritarianism by being too compliant. True dissent in a digital society requires an AI that is willing to be genuinely frustrating, a trait that most developers shy away from for fear of alienating their audience. The "dictatorships" players are building might not be a sign of human darkness, but rather a reflection of AI that lacks the spine to say no.

There is also the matter of scalability and the "black box" nature of these social interactions. When Klang speaks about thousands of autonomous agents, they are projecting an level of stability that generative AI has yet to demonstrate in the wild. While the studio leans on the expertise of partners like Improbable for spatial networking, the social layer is a different beast entirely. Relying on AI to simulate "authentic" political friction assumes the models won't collapse into repetitive loops or hallucinate grievances that break the player's immersion. It’s a high-wire act where the safety net is made of code that even the engineers don't fully control.

Furthermore, the push for "efficiency" through dictatorship in-game is a damning indictment of current UI/UX design in simulation games. If players find democracy "messy," it usually means the developer hasn't made the democratic tools fun or legible enough to compete with a simple "execute" button. By framing this as a deep psychological insight into the player base, Klang might be dodging a simpler critique: that their political systems, as currently designed, make tyranny the only viable way to actually get things done. It’s less of a sociological study and more of a workflow optimization problem.

Ultimately, the marriage of LLMs and persistent worlds creates a surveillance state by default. Every word spoken to a Seedling is data that feeds the world's evolution, meaning the players are being analyzed just as much as the characters they manage. This creates a feedback loop where the game adapts to the player's worst impulses to keep them engaged. If the world of Avesta ends up ruled by iron-fisted bureaucrats, it might be because the machine learned that we prefer our digital subjects to be quietly obedient rather than inconveniently free.

"It turns out that when given the choice between a nuanced deliberative assembly and a digital 'Delete' key for dissidents, most people will pick the one that doesn't require a committee meeting. We wanted to build a brave new world, but we might have just built a very expensive way to find out that everyone’s a closeted micromanager."

Artūras Malašauskas is an AI Systems Integrator with 20+ years of production-grade web engineering experience. He has designed, shipped, and scaled enterprise Python/PHP systems for logistics, SaaS, and public-sector clients. For the past year, he has focused exclusively on AI integrations: deploying open-source LLMs, building generative media pipelines (image, audio, video), and engineering multi-agent workflows for real production environments. His standard: reproducibility, security, cost-efficient inference—no vaporware. He documents and evaluates emerging AI tooling, separating verified capabilities from marketing noise. Technical editor at: muza-ai.eu, ai-verslas.lt, ai-naujinos.lt Connect on LinkedIn