Thanks for such a useful list! If we're meant to be looking for quality content and ignoring the bad content then we can agree that meta-data around the content (tags, author, etc) are indicatory only. Therefore, better tools will need to delve into the content itself.
FWIW, I have a philosophical objection that AI can reliably make subjective quality decisions - but AI can certainly help.
FWIW, I have a philosophical objection that AI can reliably make subjective quality decisions - but AI can certainly help.
Well I agree, but it can learn and classify the good from the bad if you give it enough quality input. But it could also do things such as grammar checking. I mean when did you use Google translate for the last time? I used it on Chinese and Spanish to English. I was pretty blown away, it was very readable, almost as if a human had written it.
Since quality is subjective on a community by community basis then it's going to be pretty rough making a one-size-fits-all Quality Control AI. And, since the AI can only learn from posts that have happened then it will always trail behind where the community is currently at.
You can already see certain online communities (e.g. 4Chan) actively work to subvert AIs.
Thanks for such a useful list! If we're meant to be looking for quality content and ignoring the bad content then we can agree that meta-data around the content (tags, author, etc) are indicatory only. Therefore, better tools will need to delve into the content itself.
FWIW, I have a philosophical objection that AI can reliably make subjective quality decisions - but AI can certainly help.
Well I agree, but it can learn and classify the good from the bad if you give it enough quality input. But it could also do things such as grammar checking. I mean when did you use Google translate for the last time? I used it on Chinese and Spanish to English. I was pretty blown away, it was very readable, almost as if a human had written it.
Since quality is subjective on a community by community basis then it's going to be pretty rough making a one-size-fits-all Quality Control AI. And, since the AI can only learn from posts that have happened then it will always trail behind where the community is currently at.
You can already see certain online communities (e.g. 4Chan) actively work to subvert AIs.