Almost every online gamer is likely to have already had experiences with the toxic behavior of other players. The spectrum ranges from trolling and insults through discrimination and marginalization to threats and stalking. In most online games, players have the option to report someone else’s behavior after a game. But despite this system, many users have the feeling that nothing has really changed in the initial situation. President J. Allen Brack spoke for the first time about Blizzard’s own AI and the successes that have been achieved in recent years.

In the so-called “Blizzard Fireside Chat 2020 “said Brack that the AI ​​has been used in the popular titles Overwatch and Heroes of the Storm for a long time and has significantly reduced insults and the like. Above all, multiple offenders would have been subject to sanctions and bans faster, which would have contributed significantly to the success.

Also in the well-known online role-playing game World of Warcraft the AI ​​is already in use. There the number of players who regularly displayed toxic behavior could be reduced by half.

The main aim of the AI ​​is to accelerate the player reports and to relieve the human employees. For this purpose, the AI ​​is fed countless texts that are displayed as offensive, discriminatory and otherwise problematic. Using algorithms, the artificial intelligence then learns to independently establish connections and, in the best case, to correctly interpret the sentences and words in the context of the chat. If the reported text is offensive neither for a player nor for the AI, it will still be forwarded to human employees, who can then better assess it.

When and whether the AI ​​too will be used in other Blizzard titles, was not revealed in the interview.