When it comes to esports, playing against human opponents usually provides the biggest thrill, thanks to their unpredictability and deep strategic planning. But sometimes, that constant challenge is exhausting—what if you want to just sit back, relax, and play a match against a bunch of voiceless drones?
In games like League of Legends, the most popular esport in the world, you can do this against what are known as bots—computer controlled opponents. But in League’s case, the bots programming leaves a lot to be desired, and they have traditionally performed so poorly they’re pretty much only useful as training fodder for newbies.
The game’s developer, Riot Games, announced yesterday that, five years after its release, League is finally getting some major upgrades to its bot AI. The main goal? To make them act more human.
The basic idea is this: by creating a more “human” bot, players get more consistent opponents that react more realistically, rather than opponents that are sometimes brilliant, but other times just plain horrible.
One key element that factored into this is something called threat assessment. The previous bot AI naturally considered damage a major threat, so a bot that was just hit by a bunch of spells would run away in fear.
But in reality, because those spells have cooldowns—periods a player must wait before using them again—surviving a sudden burst of damage means you are actually in a strong position of power. Any human in that situation would react by dishing out their own spells, not running away.
Under the game’s new AI, bots now keep track of spells and will make their own counterattacks. Compare the following behavior to the one above:
Other changes involve bots attempting to dodge spells and having them play with up-to-date item and spell builds.