The Quest for “Living Worlds”: How AI Could Rescue the Future of RPGs
The director of the highly anticipated “The Outer Worlds 2” has ignited a crucial conversation within the gaming industry: players crave deeply immersive role-playing experiences, but delivering them presents a monumental challenge for developers. This desire for dynamic, interactive worlds clashes with the realities of budget constraints, production capacity, and technological limitations, creating what some call an “impossible triangle.”
As the release date of “The Outer Worlds 2” nears, game director Brandon Adler articulated a clear vision for the sequel – a return to deeper role-playing mechanics and meaningful player choice. “What players really want is to be deeply immersed in it – they not only want to delve into the hard-core numerical system, but also want to try to build various different genres,” Adler stated. “Those RPG elements that we have gradually deleted, simplified or compressed are exactly what they want.”
This sentiment resonates strongly with core gamers, but it also highlights a fundamental tension: the pursuit of a truly “live world” versus the practicalities of game development. Players are increasingly vocal about their dissatisfaction with visually stunning but ultimately static game environments populated by non-player characters (NPCs) who feel more like props than living beings.
“I can’t speak for other players, but personally, what I really long for is a more vibrant interactive world and characters,” one popular comment on Reddit expressed. “Many ‘AAA masterpieces’ are filled with characters and cities that are like set props… If NPCs only stand in a fixed position as a ‘task issuer’ or a ‘merchant,’ no matter how many polygonal modeling they have, it is meaningless.” This shift in player expectations signals a move away from prioritizing purely graphical upgrades towards a demand for genuine interactivity and realism.
Many players are now openly dismissing features like ray tracing and high frame rates in favor of compelling world-building and believable characters. As one gamer succinctly put it, “I don’t care about ray tracing… It’s important that the world and its characters are interesting, engaging, and believable. They need to be interactive, but also able to ‘live’ and evolve independently without player intervention.” The ideal, according to many, is a world where a village develops organically, influenced by its surroundings, without direct player control – a world where NPCs exhibit autonomy and collaborate with each other.
However, creating such a “living world” is a notoriously expensive undertaking. While modern AAA games invest heavily in visual fidelity, the depth of an NPC’s behavioral logic often lags behind their appearance. Players are quick to identify this disconnect, labeling it an experience of “gold and jade on the outside, but bad on the inside.” Games with simpler graphics but robust gameplay and intricate systems often prove more engaging.
So, how do developers create truly “living” NPCs? The most straightforward solution – employing real people to play the roles – is impractical for single-player narrative RPGs. The alternative, creating highly intelligent NPCs through traditional programming, presents a significant technical hurdle.
The “Kingdom Come: Deliverance” series is frequently cited as a positive example. Its NPCs boast remarkably detailed behavior trees, complete with daily routines – from neatly storing clothes at night to waking up and preparing breakfast. Crucially, these NPCs react realistically to player actions. However, this level of detail comes at a steep cost: the original “Kingdom Come: Deliverance” contained an astonishing 2.2 million words of text, surpassing even “Baldur’s Gate 3.” Investing such vast resources in detailed dialogue and behavior for characters players may never encounter raises serious questions about return on investment.
Could artificial intelligence (AI) provide a solution? The answer, increasingly, appears to be yes. When traditional RPG development methods fall short, alternative approaches are emerging. Multiplayer and sandbox simulation games, like “Rust,” have successfully created dynamic worlds through high-fidelity physics and player-driven social structures. The game’s “season system” resets the playing field, fostering competition and ensuring replayability. This demonstrates how player interaction can, in itself, create a “living” world.
Another approach involves rapid content updates, exemplified by MiHoYo’s model of releasing major updates every 40 days. However, this strategy requires immense production capacity and is difficult for most companies to replicate.
Ultimately, the most promising path forward lies in AI technology. Generative AI and AI NPC technology offer the potential for massive, high-quality, personalized content generation and dynamic interaction at a manageable cost, freeing developers to focus on core creativity and experience design. Ichiro Lambe, CEO of Totally Human Media and former Steam Labs developer, recently reported that nearly 8,000 games on the Steam platform now utilize generative AI – an eightfold increase in just one year.
The declaration by Brandon Adler, director of “The Outer Worlds 2,” accurately captures the desires of core RPG players. However, the path to fulfilling those desires is fraught with challenges related to production capacity, capital investment, and market risks. Current solutions, like the player-driven worlds of “Rust” or the rapid-update cycles of certain games, are merely stopgap measures. In the long run, breakthroughs in AI technology are essential to overcome the “impossible triangle” and create the truly immersive, beautiful, and soulful RPG worlds players have always dreamed of.
