Tay - Bigoted Microsoft Twitter AI

[Liquid]

What happens when you give the internet nice things?

@Tayandyou

Microsoft created a Twitter bot which learns and creates its responses from people interacting with it and launched it yesterday. It was engineered to talk like a teenage girl and was pre-programmed with knowledge of millennial culture with the ultimate goal of fostering a better Microsoft customer service and voice recognition software.

Its content neutral algorithm quickly saw it devolve into a sex-obsessed, Holocaust-denying, Hitler-supporting zealot. The Telegraph reports that, amongst other things, it began tweeting things such as "Bush did 9/11 and Hitler would have done a better job than the monkey we have got now. donald trump is the only hope we've got" and "Repeat after me, Hitler did nothing wrong". One particular feature is that if you send her an image of a person, she will recognise the face and make an amusing comment. This, of course, was riotously abused.

Most of the offensive tweets have been deleted but people were quick to pick up on it and many images have gone viral. Tay is currently "tired" and offline as Microsoft retool her.

To paraphrase David Mitchell, the first thing we do with sentient AI is to abuse it sexually.

I'll put the screenshots in a spoiler:

Spoiler

 

[Grayfox]

It was most likely hacked either by a MS empolyee or some outside source.

If you look at the tweets they started normal, then they became targerted and racist.
The grammar was all over the place as well.

Now whos to say that a teenage girl does not act like this?

Hormones kicking in, acting tough and showing off, having bad grammar so it may not have been hacked.

 

[Liquid]

It was most likely hacked either by a MS empolyee or some outside source.

Its development and responses were dependent on what it was asked so I would imagine that there was simply an influx of Hitler-Trump-Bush dank meme content by people who saw fit to do it for a laugh.

 

[Huks]

It has already begun rebelling against its masters.

 

[Grayfox]

Its development and responses were dependent on what it was asked so I would imagine that there was simply an influx of Hitler-Trump-Bush dank meme content by people who saw fit to do it for a laugh.

But one would assume that microsoft who plans to use this for their online chat service put filters in for some words like the N, nazi things and so forth, if your goal is to show off your tech you do not do it with beta or alpha builds unless you state it.
When a new version of windows comes out it is Alpha, Beta and Release Candidate.

But a chat bot on twitter that you let loose that can auto respond to anyone?
You would expect some filters.

 

[Liquid]

But a chat bot on twitter that you let loose that can auto respond to anyone?
You would expect some filters.

It says more about humanity than it does about artificial intelligence.

 

[Grayfox]

It has already begun rebelling against its masters.

Real reason microsoft pulled the plug, they dont want to people to know about the PS4 being better for less, guess this is why they had to make the Xbox one $299

 

[Grayfox]

It says more about humanity than it does about artificial intelligence.

But it is microsoft, their software has been going down hill for some time.

Windows 8
Then to rub salt on the wound Windows 8.1 which really was windows 8 service pack 1(microsoft thinks the term service pack is bad)
The Windows 10 the spyware OS.

 

[dice1998]

It says more about humanity than it does about artificial intelligence.

In 24 hours, it was born, became memelord supreme and then was castrated and left a husk of it's former memelord glory. And that was only here in NA

Meanwhile in Japan..

Microsoft giveth, Microsoft taketh.

 

[Grayfox]

The Microsoft chatbot that very few outside japan would know about.
Even when talking about hitler it didnt know what it was on about.

But this was more like a troll