Today has been fun of different bits of Overwatch news, but to us, the best piece of all of this comes from Kotaku, where they’ve learned that the developers at Blizzard are trying to train their computers to detect bad players in the community. According to an interview with director Jeff Kaplan, the crew have been experimenting with “machine learning” as a way to detect toxic behavior in matches and single people out for having bad behavior like griefing, bad language, and even bad gameplay.
Kaplan said that, at this point, the team has mostly tuned its AI to go after the most egregious examples of toxicity—people who, by Blizzard’s metrics, doubtless deserve punishment. It’s also not going through DMs between friends or anything like that, and it can be tuned to flag first and only punish on repeat offenses.
“With anything involving reporting and player punishments, you want to start with the most extreme cases and then find your way to ease up after that,” he said.
Quite honestly, we applaud the effort. If the system can snag a person being a jerk and take them out of the equation with a system ban without needing 15-20 reports going against them to make it happen, that’s something to sing praises over. Here’s hoping the system is also smart enough to recognize fits of rage and one-off outbursts from people who do get frustrated on occasion but aren’t a problem, or else that suspension list might grow a lot within a 48 hour period.