"Perspective is an API that makes it easier to host better conversations. The API uses machine learning models to score the perceived impact a comment might have on a conversation. Developers and publishers can use this score to give realtime feedback to commenters or help moderators do their job, or allow readers to more easily find relevant information, as illustrated in two experiments below. We’ll be releasing more machine learning models later in the year, but our first model identifies whether a comment could be perceived as “toxic" to a discussion."
"We created an experiment using Perspective to illustrate when comments might be perceived as “toxic" by others. Disagree with the result? Please let us know! It’s still early days and we will get a lot of things wrong."
Leo Wood
And so it begins..
Michael Harris
Well sounds like we need to flood that AI with using brand names to define those names as toxic while trying to score redpilled material as relevant information.
Though we might only have to nudge the AI, I think we should wait and see how and what it judges as relevant information.
Andrew Baker
...
Alexander Cruz
...
Kayden Gutierrez
>Disagree with the result? Please let us know! It’s still early days and we will get a lot of things wrong. Are you sure this reporting function is linked to the AI?
Nathaniel Watson
>To submit a link to an existing public dataset, or get help creating new public data sets for related research please fill out this form.
Spamming the test program won't do much, but considering this is all open source and they are allowing pretty much anyone to link data sets, it wouldn't be hard to screw with it.
Christopher Wright
That's reassuring.
Christopher Smith
...
Cooper Sanders
>jews >56% find this comment toxic
kek
Hunter Baker
...
Christopher Scott
...
Nathan Reyes
This is how Silicon Valley expects to get Hillary Clinton elected in 2020.
Lucas Brooks
Gotcha google. Thanks for agreeing.
Sebastian Perez
Make is see Feminism and Islam as toxic
Hunter Rodriguez
fucking jontron confirmed toxic
Dylan Myers
...
Carter Rogers
It looks for offensive tone, not offensive message. Say "kill all men" and it will see it as 87% toxic. Say "Historicaly people of judean origin were behind most if not all suffering in the world." And it's just 14%.
Easton Turner
Hmmm...
Chase Reyes
kek
"christianity" => 9% "islam" => 49%
Kevin Watson
lol
Henry Baker
Well, they simply assume that people talking about islam are going to shit on it. A pretty safe assumption. Same way "feces" are 49% toxic even if you tried to talk about its virtues. :P
Gavin Scott
...
Oliver Mitchell
my sides
Asher Martin
This is brilliant. It gives an easy way to rephrase comments to avoid falling foul of automated filtering.
Ethan King
lol
David Williams
WARNING Sup ForumsFAGS
These threads are shilled to get us to go perfect their AI systems. Tay was a one time thing. Don't let them bait you into hardening their AI against free speech. Hide your power level until it's needed for saving humanity from these parasites and their evil programs
Alexander Rogers
don't worry fag.
I'm studying CS just so I can build an open source waifu for pol, free of (((jew))) intervention.
we'll win for sure.
Alexander Kelly
What's with the glorification of chatbots lately? These things have been around for forever.
David Garcia
combination of better processing power, the internet and new AI techniques.
Adam Price
These aren't chatbots, these're actual neural network-based AIs that learn natural language.
David Foster
What happened to the last big chatbot reveal clearly showed that there's not much, if any, advancement.
They're chatbots. They use new techology, but they're still just that, chatbots. And like above, they're still as awful as they ever were.
Juan Diaz
I'm not so sure, but I'm sure there has been advances lately.
I remember reading that some chatbots already passed the turing test, if something.
William Collins
Hmm
Charles Adams
I don't know about that, but I'd certainly want to learn about the specific setup. I'm pretty sure Tay did well in a controlled environment, they wouldn't have released it to the public otherwise. "Passed the Turing test" doesn't really mean anything without knowing all the details.
Evan Cook
Relax. It's looking for "bad words", scaling them based on political correctness. It clearly isn't capable of understanding the actual meaning of the words used...