Learning from Tay’s introduction • Hacker News Board

Re: Learning from Tay’s introduction

#181

Post by twoSeats » Sat, Mar 26, 2016, 9:00 AM UTC

bliti wrote: post #81

dsp1234 wrote: post #50

Earlier quoted context omitted.

because we had thought about all likely scenarios in our comfortable offices. Or because they had real world experience with another bot in a similar situation: "In China, our XiaoIce chatbot is being used by some 40 million people, delighting with its stories and conversations."

Interesting that it worked in China and not here. I'd like to read about the difference rather than a bland apology.

Where exactly is "here" though? Twitter is global remember? Anyone in the world, including Zambia where I am could have interacted with Tay. That's a very different proposition to "China".

Re: Learning from Tay’s introduction

#182

Post by vidarh » Sat, Mar 26, 2016, 9:00 AM UTC

JabavuAdams wrote: post #141

Houshalter wrote: post #121

Microsoft please don't worry about this. No one but idiots are offended by this. It's understood that its just a stupid chatbot mimicking human responses. The AI isn't terrible, people are. And unless you keep a dataset of every offensive thing a person can say, and every offensive image they can tweet, there's no way to prevent people from tweeting it pictures of Hitler... or Scunthorpe. But who cares. This is just…

They should worry about this, but not because of the offense and outrage (mock or real -- I don't know). I'm certainly not offended, just amused. It just seems kind of obvious that Twitter users would grief the bot. Now, they address this point, but it's interesting that they still didn't last a day. Think about that -- they were expecting abuse, but they still lasted less than 24 hours. That's certainly interesting.

It is also an illustration of a problem with AI that trusts it environment and how such trust could have really bad consequences if said AI is allowed to anything important.

People try to social engineer people all the time, and it works. The consequences are somewhat limited because you are limited to breadth (mass media, phishing etc.) or depth (one on one interactions), but can still be scary.

If an AI isn't more resistant, we face the risk that any re-purposing of data from one AI in other copies, or allowing the AI to massively multitask makes social-engineering of AI far more wide-reaching, and hence far more worthwhile for attackers.

It'll be interesting to see what kinds of attacks will get directed at e.g. customer service bots in the future, and what brands ends up damaged as a result.

Re: Learning from Tay’s introduction

#183

Post by ivoras » Sat, Mar 26, 2016, 9:09 AM UTC

When they say "exploited", is it actually some sequence of words which was interpreted by Tay as learning commands, or was it simply repeating to it "Hitler is love" a thousand times? Any records of how it learned?

Re: Learning from Tay’s introduction

#184

Post by SixSigma » Sat, Mar 26, 2016, 10:08 AM UTC

zxcvvcxz wrote: post #109

> The great experience with XiaoIce led us to wonder: Would an AI like this be just as captivating in a radically different cultural environment? Tay – a chatbot created for 18- to 24- year-olds in the U.S. for entertainment purposes – is our first attempt to answer this question. Maybe you did answer that question. No, the answer is not that 18-24 year olds in the US are racist and what not. But that the ones respon…

Are you sure it was hateful content?

Is Mel Brookes [1] singing "Springtime for Hitler" spreading hateful content.

I think the word you are looking for is "irreverent".

[1] http://i.dailymail.co.uk/i/pix/2009/04/09/article-1168749-00...

Re: Learning from Tay’s introduction

#185

Post by philip142au » Sat, Mar 26, 2016, 10:34 AM UTC

Why didn't they just parse the output of Tay and map the meaning of the words against Wordnet and filter for things which had negative meanings?

Re: Learning from Tay’s introduction

#186

Post by philip142au » Sat, Mar 26, 2016, 10:36 AM UTC

If a human said the things Tay said on twitter, the police would be all over him. Why isn't the police then logically all over Microsoft?

Re: Learning from Tay’s introduction

#187

Post by greyman » Sat, Mar 26, 2016, 11:10 AM UTC

pervycreeper wrote: post #152

bliti wrote: post #13

This reads like science fiction: Learning from Somebot's introduction. Last week we deployed Somebot to $location. Our confidence in Somebot was high because we had thought about all likely scenarios in our comfortable offices. We are deeply sorry and assume the full responsibility from Somebot's actions. Moving forward we will make sure to include safeguards to reduce the amount of pain caused by Somebot's deploymen…

The response that Microsoft gave was really the worst of both worlds, combining the systematic corporate shirking of responsibility which enables and gives cover to a great deal of evil, with a complete accedence to and embracing of political correctness. Furthermore, their claim that they did not anticipate "this specific attack" is either a lie, or their creators made an extremely obvious mistake (yes, easy to say…

IMHO the response from Microsoft was fine. They admitted the mistake and taken the bot offline. What else should they do? What I find strange is that nobody criticizes twitter for not doing anything... I think Facebook would deal better with this.

Re: Learning from Tay’s introduction

#188

Post by douche » Sat, Mar 26, 2016, 11:45 AM UTC

tariqali34 wrote: post #174

mtgx wrote: post #172

Earlier quoted context omitted.

Come on. If your AI can't handle the Internet then it doesn't deserve to be on the Internet, let alone in charge of important stuff.

Nah, it can handle the Internet just fine. It just can't speak on the Internet.

It appears to have learned how to speak Internet pretty alright.

Hopefully this gets integrated right into Cortana. "Cortana, how do I get to the hardware store?" "Turn left and yourself"

Re: Learning from Tay’s introduction

#189

Post by douche » Sat, Mar 26, 2016, 12:03 PM UTC

philip142au wrote: post #186

If a human said the things Tay said on twitter, the police would be all over him. Why isn't the police then logically all over Microsoft?

Why would the police be after you for saying dumb shit on the internet? Unless you're posting child porn, anyway.

Note, that the US doesn't have any real (stupid) hate speech laws.

Re: Learning from Tay’s introduction

#190

Post by lostcolony » Sat, Mar 26, 2016, 12:16 PM UTC

throw11231 wrote: post #180

azakai wrote: post #96

Earlier quoted context omitted.

P.S. And we would be remiss if we did not note that they did, in fact, have 20 seconds to comply.

"he did". In the fine movie a single male test person was chosen. Sorry for mentioning this, but recently HN is unreadable due to PC use of pronouns.

But in referencing those affected by Tay, it was both genders, hence their usage of it. This was not 'PC use of pronouns', but 'correct use of pronouns', due to said pronoun's use in drawing an analogy between Microsoft's response and that movie scene.

You're a little alarmist. That's also quite a bit of hyperbole, that PC use of pronouns leads to something being -unreadable-. But you know that, hence your use of a throwaway account.