San Francisco-based nonprofit AI research organization OpenAI has launched Neural MMO, a "massively multiagent" platform for training AI in complex, open-world environments.
In the Neural MMO platform, AI agents spawn randomly in a virtual training ground – which takes the form of an RPG-like world – in automatically generated tile maps. They then must navigate around the map to acquire 'food' and 'water' to keep them alive, in addition to fending off other AI characters within the gameplay who are also competing for the available resources.
There are three combat settings: 'melee', 'range' and 'mage' used to battle with other players.
OpenAI incentivized the AI system by 'rewarding' agents with a longer lifespan when they did well in the game and found that the longer they interacted with one another, the better they became at certain tasks.
"The game genre of massively multiplayer online games simulates a large ecosystem of a variable number of players competing in persistent and extensive environments," OpenAI wrote in a blog post announcing the release of the platform. "The inclusion of many agents and species leads to better exploration, divergent niche formation and greater overall competence."
In short, the research team found that training more AI agents per map always resulted in stronger performance when the servers were 'merged' and agents were competing with one another.
"The gap between the work 'real' AI researchers do and what games and game developers tend to incorporate is massive," noted Extreme Tech. "But projects like this hint at ways to one day bridge the two. Imagine playing a game where the NPC characters weren't just well-scripted, they were actually capable of learning and fighting more effectively, adapting themselves to your own methods of playing a game and working with you to achieve mission goals."