Video games have always been at the forefront of experimentation with artificial intelligence systems. AI has mainly been used to provide competition to players or develop a narrative storyline through non-player characters (NPCs).
The advent of natural language processing, strong intent engines and machine learning opens up new ways to use AI in games. One of the obvious examples is to create more lifelike AI characters that can converse with players, compete and provide a more realistic experience.
However, AI can also be used in somewhat unexpected ways to improve player experience. One of the most promising areas is to provide immediate help within the game itself.
Currently, most players need to leave the game to get support, which suspends play and dampens the experience. This also requires going to a support site to either search and read documentation, which players would rather not do, or submit a ticket to a human agent that may take time to process, which means waiting, which they also don’t like.
An in-game AI agent can give players instant help. Players can immediately resolve problems they might be having with the game, their account, or other players. Players can also get instant access to tips to help them play the game better. They can access all of this without ever leaving the game. With a natural language bot they can also avoid having to trudge through documentation, and instead simply ask the bot a question and get a direct answer.
From a game company’s perspective, this also dramatically reduces the burden on human agents, allowing them to focus on more personal-touch interactions.
Riot Games is a pioneer in using AI this way. They launched Blitzcrank Bot (named after one of the “League of Legends” heroes!) in 2015. Blitzcrank Bot was trained on a dataset of millions of support tickets submitted to Riot. After training, the bot can classify text in a support ticket to understand what the issue might be and respond or route accordingly. Blitzcrank Bot was a huge success, processing over 20% of Riot’s tickets and doing the work of over 50 agents as of June 2017. You can read about it here.
More than ever, video games provide fertile ground to use AI to improve player experiences in a wide number of ways, both in gameplay and in the broader game experience.
We’ve realized that our approach to AI characters for entertainment can also be applied to providing in-game player support. By creating intelligent natural language support agents, we can have a greatly expanded positive impact on player experience across a larger number of games.