On: Microsoft chatbot is taught to swear on Twitter

[golfer]

Google's AI beats go champions. Microsoft's AI turns into a racist genocidal maniac.

[sremani → golfer]

Tay is more a reflection of the interwebs of today, than the culture or values of Microsoft. I think we should be cautious about our conclusions.

[golfer → sremani]

Perhaps, but it is naive of the Microsoft researchers to think this wasn't a possibility. They should have seen this coming and prepared accordingly.

[Crito → golfer]

At worst, Microsoft researchers are guilty of not being familiar with internet racism. Hardly a great sin.

[siegecraft → Crito]

They didn't sanitize their input data.. that's the worst sin you can commit.

[Crito → siegecraft]

That's pretty fucking hyperbolic. A technical "sin" perhaps, but people are heaping derision on them as though they committed some great moral sin.

[siegecraft → Crito]

Hrm, I assumed it would be obvious I meant a technical sin

[794CD01 → golfer]

Google's AI was designed to play go, and became an expert at Go. Microsoft's AI was designed to use twitter, and became an expert at twitter.

[umeshunni → golfer]

Add this to the manual on how not to do a PR stunt

[Kristine1975 → umeshunni]

Everybody's talking about it, aren't they? "No such thing as bad PR" an all that...

[Grishnakh → golfer]

The next time someone talks about purchasing or using Microsoft software, you can point to this and ask them if they want to support a company that literally agrees with Hitler, and has said so publicly!