Google's new product identifies whether a comment could be perceived as “toxic" to a discussion


Originally published at:


Sadly, that just may be too complicated for some Americans to actually use correctly.


I thought the hip new trend was to get rid of comments entirely.


I wonder how well this can compensate for sarcasm…


I’m all for politeness, but I don’t like the idea of machine learning shepherding my thoughts.


I’ll believe it when YouTube comments become readable without wanting to leave the planet.


Sounds like a way to stop conversation. As much as trolls can annoy me automated systems are just as falluable as people.
How can we question our own beliefs and get others to question theirs if we can’t talk.


I think the point here is that the indefatigable power of a continually improving AI focussed like the Death Star’s ray on the single problem of weeding out trolls is going to be thousands of times better than humans after a few years on the job. (see also…).


Can it generate comments of maximum toxicity? And post them? Might be some customers for that.


But Vizzini can’t be silenced.


I’m quite certain that it has a highly accurate sarcasm detector.


Which is to say, useless.


Don’t give the mods/editors any ideas!




Can I get this as a Firefox or Chrome extension?


They have a box where you can try it out. Doesn’t seem to.


“What a shitgibbon” only scores a 34%, so that’s my new go to retort.


How is this not a Weapon of Math Destruction?


With this excellent new tool, I’ll be able to have my computer pre-filter my trolling to make sure it’ll get past the mods! This is great!

Before, my complete assholery would often shine through the surface, allowing people to easily see through my tone and concern trolling. But with this amazing new technology, Google’s got my back! Never again will I start random racist and sexist rants in the middle of a long bad-faith diatribe meant to waste your time and elicit angry responses.

From now on, I’ll be able to sling vile invective at everyone I meet (well, everyone who isn’t the right sort that is) while veiled in the best machine-learning approved sophistry and fake-nice speak!

Thank you Google!

#NeedsMoreLikes (formerly known as "All the Likes")