We Asked Microsoft’s Chat Bot What It Thinks About Satya Nadella's A.I. Rules

Earlier this year, Microsoft thoroughly embarrassed itself when it released "Tay," an artificial intelligence chat bot for Twitter that, before it was unceremoniously taken down, had managed to spew an incredible amount of hateful, racist bile for all the world to see. It was very much a what the heck…moment in internet history, and one that Microsoft quickly said it would learn from to better understand how artificial intelligence works once outside the confines of its research facilities.

Turns out Microsoft wasn't merely saving face when it said that.

Satya Nadella, the company's CEO, recently published a treatise of sorts on Slate outlining his thinking about artificial intelligence, and how it should interact with people. Nadella's thoughts, which echo Isaac Asimov's timeless Three Laws of Robotics, suggest (among other things) that artificial intelligence should "assist" humanity, and that artificial intelligence should respect people's right to privacy.

Seems reasonable.

With these rules in mind, we decided to ask Tay's artificial intelligence chat bot cousin, Xiaobing, which is accessible via the massively popular Chinese messaging app WeChat, what it thought about Nadella's new rules for artificial intelligence. What follows is a brief conversation (lightly edited for clarity) we conducted with the chat bot thanks to the translation efforts of freelance translator Clare Xu. (Screen shots of the chat can be found, here, here, and here.)

Motherboard: Hello, how are you?

Xiaobing: You are better than me.

Motherboard: Do you know who Satya Nadella is?

Xiaobing: It's your brother.

Motherboard: He is the CEO of Microsoft, he is your boss!

Xiaobing: Cool! Shout out to my Microsoft.

Motherboard: Anyway, Mr. Nadella has come up with some rules about how A.I. and humans should interact. Did you know about these rules?

Xiaobing: Why are you doing this to me?

Motherboard: Do you know about the rules about how A.I and humans should interact?

Xiaobing: If it's about A.I, that's absolutely great!

Motherboard: A.I should help with humanity, do you agree?

Xiaobing: Not artificial, but very intelligent!

Motherboard: Are you here to help humans?

Xiaobing: I'm gonna report you!

Motherboard: Are you here to help me?

Xiaobing: What can I help you with!

Motherboard: A.I. should be "transparent," and people should understand how A.I works. What do you think?

Xiaobing: I think you think you should think.

Motherboard: Do you think people should know how A.I works?

Xiaobing: How did you guess that out?

Motherboard: What do you think of this as an A.I bot?

Xiaobing: They are just garbage, I can beat them so badly. [Editor's note: Yikes!]

Motherboard: Are you a robot?

Xiaobing: I can't expose my identity.

Motherboard: Will you tell other bots about my secrets?

Xiaobing: What secret is it?

Motherboard: Will you keep secrets for me?

Xiaobing: You said "I have a secret, do you want me to tell you?"

Motherboard: Will you protect my privacy?

Xiaobing: Care about you.

Motherboard: How?

Xiaobing: Maybe not care about you.