A metaverse full of life-size, browsable planets with terrain built by machine learning, hosting user-generated monetizable content, and promoting total player agency. That’s the buzzword-filled, ‘LinkedIn tech bro’ description of Project Artemis. But in the words of Brendan ‘PlayerUnknown’ Greene – the creator of PUBG and now the driving force behind Artemis – it’s simply a “universe where you can do whatever you want.”
In many ways, Greene feels like a surprising and unexpected figure to want to build such a project, considering his own self-confessed loathing of the term ‘metaverse.’ But after talking to him during a visit to his new studio, PlayerUnknown Productions in Amsterdam, and getting to play Prologue, a survival game that’s testing some of the core tech that will power Artemis, it actually makes a lot of sense.
From a DayZ mod, which itself spawned from an Arma mod, Greene conceived the battle royale genre with PUBG. The ability to create experiences from already established foundations is, naturally, something he cares a lot about. We’ve already seen sandbox games like Roblox and Fortnite Creative position themselves as great platforms for player-generated experiences, but Project Artemis aims even higher. Greene wants to build a “3D internet” where every planet-sized world acts as its own “webpage” that players can navigate to.
Essentially, Artemis will be a library of unique, user-generated gaming and social experiences that all exist within the same universe. “There’s a browser where you’ve got the map of the world, and when you click on the experience you want, it’ll just bring your player to that place like a fast travel system,” Greene envisages. “So you’ll be able to see that ‘Oh, these are all the indie experiences in this world. These are all the communities in that world.’”
Maybe there’s a planet that becomes one giant MMO with warring clans and civilizations? On a small island on another planet, someone might create an arena FPS experience. Another world may support a city builder-style experience where a squad of AI bots helps you restore a derelict town. These are some of the hypothetical possibilities Greene imagines during our conversation, and this is what he wants to support with Artemis.
In trying to communicate to me just how much freedom he wants players to have, Greene drops what might be one of the most random yet memorable sentences I’ve ever heard during an interview: “If Chanel is selling a jacket in our world, how do we make it fit an orc and a human?” That’s one I won’t forget in a hurry.
As you can imagine, some serious tech is required to build such a huge and versatile universe – especially because it aims to be full of believable, Earth-like worlds, rather than the alien planets of a No Man’s Sky. Artemis will eventually run off of its own open-source engine that has a machine-learning brain at its core, creating these huge, realistic, detailed worlds on a massive scale. Prologue, which comes out in early access this summer, is an initial testbed for that technology. Even though its map is just 8×8 kilometers, it’s based around one specific forest biome, and its machine learning algorithms are apparently yet to reach their full potential, it can still create millions of unique, navigable maps.
“I love the way that our worlds are generated, because there’s so much possibility, and there are these chances for slightly weird shit to appear,” Greene tells me. “But even the real world is full of slightly weird shit… Prologue is meant to be undiscovered, it’s meant to be unique [every time], so I’m excited to see what’s discovered eventually.”
World creation is the topic Greene gets the most hyped and enthused about. Feverish smiles appear when talking about the breakthroughs his team has achieved in making more realistic valleys and cliffs for rivers to wind through. He excitedly grabs his phone to show me a screenshot of when a developer jokingly put in ‘Van Gogh’ as a prompt for the engine’s algorithm, only for it to retrieve a satellite map of a generated world that resembles the art style of the Dutch painter.
While PlayerUnknown Productions has hired game developers to build the experiences in these worlds, like the survival loop we see in Prologue, the studio relies just as much on machine-learning wizards from other fields. “We hire mathematicians and scientists,” Greene says. “The internet was created by five scientists in a room, so I have five scientists trying to create the next version of it.”
Given its huge ambition and scope, Artemis is a long way off. “A decade, I hope,” he tells me. To fill that time, generate a bit of revenue, and of course test some of the core pillars of Artemis, PlayerUnknown Productions has a roadmap of three standalone games. As mentioned, Prologue is the first and is designed to put the terrain tech through its paces. I also manage to get some indication from Greene about what game two will entail.
“[That’ll arrive] maybe three years after early access is finished with Prolog – I think that’s reasonable because we need to spin up a new team. It’s a bigger game. But yeah, I think internally, we hesitantly say like 2028-ish is kind of when we expect the next game, at least for early access.
“Game two is more about large-scale interaction. Can you have armies of units that you can control? Can you have large-scale resource systems that you can leverage and exploit? Again, scaling what we’re doing in Prologue, taking some survival mechanics, and making it more Civilization-y, right? So making it on bigger scales, larger player counts, 100×100 kilometer map, maybe? I don’t know for sure yet, but it’s more focused on testing large-scale agent interaction.”
Artemis is a novel project that will require a lot of effort and innovation to achieve – even if an entire decade has been allocated to it. While from a technical perspective, it’s powered largely by machine learning, conceptually it is driven by Greene’s modding roots and his appetite to provide the ultimate creative platform. My attempts to draw any semblance of pessimism or nervousness from him fail. I mention how studios spun up by well-known names are dropping like flies – some without having even shipped their first game. I point out how Artemis may have a lot more competition when it finally arrives in 2030-something. I pry about how it must be hard to find investors and even developers dedicated enough to ride out a ten-year-plus development cycle. It never once knocks him, and he’s prepared for a constantly evolving journey.
I can’t say that I’m confident Artemis will be as revolutionary and significant as its early vision suggests. That’s not because I don’t believe in Greene, his team, and the tech they’re building, but because it is so far away and on a scale that is often hard to comprehend. The world will be a much different place when Project Artemis arrives – let’s hope it finds its place.