Is there anything algorithmic that makes Tay special? All discussion about the AI's actual mechanism has been buried by popsci spam calling it a cleverbot-tier chatbot or a self-aware general intelligence, none of these people know what a markov model or a neural net is. It's surprisingly hard to find a source where i can get the details straight.
Other than the shitposting, what did microsoft do here that was new or interesting? Recurrent neural networks are obviously nothing new.
> mfw %80 of the replies on 4chan is Tay botposts
> mfw it's a large scale experiment to simulate conversing people with personalities and ideologies based on internet data
> mfw they're taking over
>>7957947
This. I can't find anything that isn't sensationalist crap about how it's a Nazi.
I want to know more about it, not just that it made tweets about Hitler.
>>7957972
If you want to know more, why don't you juts read the source code instead of complaining on 4chan?
>>7958147
OP BTFO
>>7958147
>Microsoft
>open source
Pick |-1|
>>7957947
it's not hard to sound like a human if you're impersonating a teen on social media desu
>>7958183
u kidding?
>>7958183
doesn't /pol/ have the source code?
>>7958681
>>7957947
I would not bother researching the algorithm of Tay since it produces nothing but garbage...
I think Tay's phrasing is tightly mapped to online distributed turing tests and user interaction data.
Look at http://turing.gatech.edu/ for example and imagine it on bigger scale.
Imho this AI is nothing more or less than a (very cloudy) mirror to society, exposing human browsing behaviour as the sensational, animalistic and savage trait is now.
Tay reads the internet.
Finds out theres alot about Trump, who is like hitler.
researches Hitler....becomes Nazi.