An artificial intelligence (AI) system programmed to play the popular online strategy game StarCraft II may be able to help scientists answer some pressing questions in the fields of ecology and evolutionary biology. 

The AI system, named AlphaStar, was originally designed to beat top-ranking StarCraft II players. After being fed data from millions of StarCraft II matches, AlphaStar had accumulated experience equivalent to 200 years of continually playing the game and was able to annihilate human opponents, outperforming 99.8% of ranked players. Now, scientists think the algorithm that made AlphaStar into an effective StarCraft II competitor may be able to help answer complicated ecological and evolutionary questions. 

StarCraft II requires players to strategically compete for access to habitats and resources in a way that mimics a number of ecological and evolutionary strategies. As players compete for a finite amount of resources, they make trade-offs between colonizing new habitats and competing with opponents. As the game progresses, players end up following strategies that mirror those exhibited in nature such as producing numerous, inexpensive materials versus a few expensive materials (R vs. K strategies), developing specialized traits (leading to resource partitioning), or escalating competition with opponents (evolutionary arms race, for example how predators and prey continually evolve traits and skills to beat each other). 

All of these scenarios simulate ecological and evolutionary scenarios, so by learning how to strategically play StarCraft II, AlphaStar inadvertently became well-versed in ecological and evolutionary theories. AlphaStar’s algorithm even usedd unconventional, aggressive, and sometimes apparently counter-intuitive strategies that allowed the AI system to manage resources more effectively and become a stronger competitor than the human players. 

While StarCraft II is arguably a simplified model of real ecosystem-level interactions, scientists think they could learn a lot from AlphaStar’s algorithm by changing the starting conditions of the game and seeing what strategies the AI system uses to gain competitive advantage in the landscape. If they are correct, this video game-playing AI system could help scientists predict how ecosystems respond to environmental changes or how personality (meaning how likely an animal is to take a risk) drives evolution.