all 6 comments

[–]d3rr 8 insightful - 1 fun8 insightful - 0 fun9 insightful - 1 fun -  (0 children)

Hell of a post, thanks for sharing it here.

[–]magnora7 8 insightful - 1 fun8 insightful - 0 fun9 insightful - 1 fun -  (0 children)

Image 2 basically says it all. Thanks for this, thought-provoking stuff!

[–]Jesus 4 insightful - 1 fun4 insightful - 0 fun5 insightful - 1 fun -  (0 children)


[–]useless_aether 4 insightful - 1 fun4 insightful - 0 fun5 insightful - 1 fun -  (0 children)

would be nice to mistrain it somehow , so its unusable. like w tay

[–]iufewal 2 insightful - 2 fun2 insightful - 1 fun3 insightful - 2 fun -  (1 child)

My first reaction was that what they were doing was completely disgusting, and should be stopped at all costs.

After having sat down for a minute and thought about it, I realized there may be alternative applications for the same model which they have developed. Currently, what the ADL has (or claims to be developing) is a programme which can pre-emptively sort speech into categories of "hate speech" and "not hate speech".

But who's to say that this programme can only be used for that purpose? Fundamentally, what they really have is a programme that sorts speech into "Category A" and "Not Category A" based on feedback from an actual team of humans.

While the ADL intends to use this technology to block what they define as "hate speech", we could in fact use the same principles to make this "category A" whatever we wanted it to be, and decide what we wish to do with the "category A" comments afterwards. For instance, if we all happened to like "movie B", we could use this programme to find all comments praising "movie B" and use another programme to instantly upvote all of them. This could be applied in advertising and so on, but it could also be used for politics; if you were partial to a particular political ideology, you could again use this to automatically upvote all the comments you agree with in any thread without ever reading a single one of them. You could also, if you were partial to a non-mainstream ideology such as Anti-Vax, immediately create a bubble around your ideology and filter out all comments that disagree with you.

If I understood the function of this machine-learning model correctly, of course.

[–]Alduin 6 insightful - 1 fun6 insightful - 0 fun7 insightful - 1 fun -  (0 children)

Man, future internet is going to suck. I miss the wild west internet days.