Twitch News & Update: Site Launches AutoMod Tool That Learns How to Make Chat Less Toxic
In what seems to be their latest concerted effort in combating chat toxicity, Twitch has launched AutoMod. This new tool makes use of "machine learning and natural language processing" to help streamers better moderate their stream chat without requiring unmanageable human intervention.
According to Twitch, AutoMod does more than just block the use of certain words on the chat. It can also detect emotes and other characters that may be used to circumvent filtering, with streamers being able to adjust it to make the chat moderation either stricter or more lenient.
The general opinion of Twitch users on stream chat is pretty low due to pervasive toxicity and abuse. Moderators have been crucial in policing chat, especially for stream channels with high viewership numbers, wherein things can get especially chaotic when subscriber-only mode isn't turned on. On the other hand, there have also been instances of mods being overly abusive with their power, banning users for little to no good reason.
Twitch is touting AutoMod as the shield to go with the sword that is the moderators. Development and testing for AutoMod has been going on for months. It recently underwent trial runs during the US election season in the Democratic and Republican National Convention livestreams, which "were commended for the civil discourse in chat." The newly-launched service shall undergo continual review, monitoring its efficacy in both site-wide and per-channel implementation.
Aside for being a moderation tool, Twitch is also putting forth AutoMod as a major part of its initiatives to promote tolerance and diversity online. While it's only available for English at the moment, AutoMod for other languages are undergoing beta testing-namely Arabic, Czech, French, German, Italian, Japanese, Korean, Polish, Portuguese, Russian, Spanish and Turkish.
While it may not be perfect right now, AutoMod has the potential to become a lot better over time as Twitch continues to develop it while listening to user feedback. The feature is now live and streamers can now activate it through the Twitch Settings page.