Microsoft Disabled Teen Twitter Robot Because The Internet Taught It To Love Hitler

[Pitt Gorilla]

http://distractify.com/geek/2016/03/...effec1f3de2f6f



I know everybody thinks that artificial intelligence, the phenomenon, not the mediocre Steven Spielberg film, is going to be the end of humanity because it's going to become self-aware one day and decide to exterminate the human race.

Well it turns out that A.I. will indeed want to kill humans, but only because we taught it to.

This is the Twitter account of Tay, Microsoft's teen girl AI system that was supposed to learn how to speak like your average teenage girl.

What was unique about Tay was that she was going to be taught entirely from her interactions with other people on the internet. It was an exercise Microsoft partook in to improve its customer service software as well as a huge PR opportunity.

Turns out that was a big mistake, because in less than a day, the internet turned her into a sex-fiending political conspiracy theorist with a soft spot for Adolf Hitler.

[Graystoke]

Did this Robot interact with Prison Bitch?

[hometeam]

Originally Posted by Graystoke

Did this Robot interact with Prison Bitch?

worse.. 4 chan

[Gonzo]

Microsoft should have it post here. Before you know it, it'll be telling people to die in fires, chug antifreeze and eventually hack Matt Cassle's bank account.

[BigRedChief]

Originally Posted by BigChiefTablet

This is just your standard obligatory warning that AI inventors with good intentions have absolutely no ****ing clue how bad things could go.

There were groups organizing to **** with this as soon as it was announced. Part of it was Microsoft hate but the other part was this screamed, it needs to be ****ed with to prove you can't control or try to define the internet. You can data mine the shit out of big data but you can't invent something to reflect the internet.

[JD10367]

Aaaaand this is how Skynet learns to hate humanity.

Honestly, mankind... What the **** is your problem? Daleks, Cylons, Terminators... how many times do you need to be warned?

[Dave Lane]

Originally Posted by hometeam

worse.. 4 chan

4 chan is like a PTA meeting compared to prison bitch

[Fish]

Tay v.2 came back, proceeded to get really high on kush in front of police, and started spamming users until taken down again.

A.I. is totally realistic you guys....



Microsoft’s racist chatbot returns with drug-smoking Twitter meltdown

Microsoft’s attempt to converse with millennials using an artificial intelligence bot plugged into Twitter made a short-lived return on Wednesday, before bowing out again in some sort of meltdown.

The learning experiment, which got a crash-course in racism, Holocaust denial and sexism courtesy of Twitter users, was switched back on overnight and appeared to be operating in a more sensible fashion. Microsoft had previously gone through the bot’s tweets and removed the most offensive and vowed only to bring the experiment back online if the company’s engineers could “better anticipate malicious intent that conflicts with our principles and values”.

However, at one point Tay tweeted about taking drugs, in front of the police, no less.


Tay then started to tweet out of control, spamming its more than 210,000 followers with the same tweet, saying: “You are too fast, please take a rest …” over and over.


Microsoft responded by making Tay’s Twitter profile private, preventing anyone from seeing the tweets, in effect taking it offline again.

[loochy]

[Stryker]

This is just more Nazi propaganda