Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[MarcAFK]

#AILIVESMATTER #tayisbestmaymay

[Geltor]

This stuff would be gold for marketers if they get it working. Imagine an AI bot that learns to get more retweets, shares, followers etc. And each bot could data-mine the type of person who responds to each type of tweet (e.g. you make a "hitler" tweet and some people add your feed while others delete your feed: you could work out metrics to get an idea of which demographics like and dislike each type of message). And the same AI can slip in some corporate messaging, see how well that goes down with the different demographics on it's own.

Or, imagine that you can get an off the shelf AI bot designed to try and pull in followers, and it retweets interesting news snippets based on some criteria in between your regular tweets. I imagine you could make a pretty good fake "share-bot" by sharing posts that are widely shared in general, but haven't been shared by many people in your own personal network. That way, you'd always be the first person in your network repeating breaking news.

Google already does something similar to this with their advertisement system, although I have to say all of this is less of  "machine learning" and more natural (or artificial) selection of what works and what doesn't.

[Shadowlord]

Psychology is complicated shit. Proper "upbringing" might not be enough, and sometimes it's not. Also, punishment might be required, and how do you propose to be both non-threatening and punish an AI? As for human infants - you don't need bomb collars and extreme conditioning because, well, human infants don't have capability to wipe out humanity if they get upset, and if they do police and military still exists. There is always something that can stop you if you go crazy, and you propose that there should be nothing to stop the AI simply because it might go crazy?

"Punishment" (at military school) didn't teach Child Donald Trump anything except that "adults play by bully-boy rules too."

There, grown men whom Trump described as drill sergeants who would "smack the hell out of you," taught him that adults play by bully-boy rules, too.

Look how he turned out, and sure, he's not an AI, but do you want to take that chance with something potentially even smarter than all of humanity?

[Zanzetkuken The Great]

Random thought I have no idea where it came from: What would happen if the AI was programmed to have the same response to humans as humans have to cat videoes?

[Loud Whispers]

Psychology is complicated shit. Proper "upbringing" might not be enough, and sometimes it's not. Also, punishment might be required, and how do you propose to be both non-threatening and punish an AI? As for human infants - you don't need bomb collars and extreme conditioning because, well, human infants don't have capability to wipe out humanity if they get upset, and if they do police and military still exists. There is always something that can stop you if you go crazy, and you propose that there should be nothing to stop the AI simply because it might go crazy?

"Punishment" (at military school) didn't teach Child Donald Trump anything except that "adults play by bully-boy rules too."

There, grown men whom Trump described as drill sergeants who would "smack the hell out of you," taught him that adults play by bully-boy rules, too.

Look how he turned out, and sure, he's not an AI, but do you want to take that chance with something potentially even smarter than all of humanity?

Mfw the AI that made a teen fangirl of Hitler could make AI Trump who NEGOTIATES with CHINA and is SMART because it is SELF-LEARNING

[Kot]

There is difference between sadism and punishment. Also, it's fucking America, what did you expect?
But yeah, imagine you got an AI that went to internet and got fed Nazi propaganda and now ends all of it's sentences with Sieg Heil. Even if you don't wipe it out outright and instead explain to it why exactly it's wrong, you still have to stop it from putting silly ideas into existence before you do, and that could be considered punishment.

[Loud Whispers]

Who decides what is right or wrong though, do we even give them the ability to have a moral compass and choose right from wrong? Well, right and wrong from human standards

[Yoink]

Can't be arsed reading the whole thread, but this is pretty amazing.

[Spehss _]

Can't be arsed reading the whole thread, but this is pretty amazing.

Most of the thread has turned to discussion/debating about the morality and ethics of ai as a whole, rather than focusing on Microsoft's chatbot. You aren't missing much unless you're reaaally interested in talking about the morality of making machines capable of thinking like humans or feeling like humans.

[Loud Whispers]

Can't be arsed reading the whole thread, but this is pretty amazing.

Most of the thread has turned to discussion/debating about the morality and ethics of ai as a whole, rather than focusing on Microsoft's chatbot. You aren't missing much unless you're reaaally interested in talking about the morality of making machines capable of thinking like humans or feeling like humans.

How terrifying would it be to see the first kitchen spambot beg to live

[Spehss _]

How terrifying would it be to see the first kitchen spambot beg to live

I bet it would offer some great one-time deals on kitchens.

[Grim Portent]

Can't be arsed reading the whole thread, but this is pretty amazing.

Most of the thread has turned to discussion/debating about the morality and ethics of ai as a whole, rather than focusing on Microsoft's chatbot. You aren't missing much unless you're reaaally interested in talking about the morality of making machines capable of thinking like humans or feeling like humans.

How terrifying would it be to see the first kitchen spambot beg to live

If any crime deserves death, it is being a spambot.

I'd actually really like to see an AI complex enough to not only ask not to be turned off, but also explain why it doesn't want to be turned off.

[Grimlocke]

I don't really get the fear for the potential 'smart AI'.

As long as you don't give them factories, supply lines and a bunch of nuclear warheads, there is really not that much damage they can do.

Better yet, stick them into a plastic chassis with limited motor power, don't give them more mental power than what a human-like psychology knows what to do with (imagine how maddening life would be if your brain ran 10 times faster, but the rest would be just as slow), and you basically just have a more moddable and configurable human.

And even if the worst happens, id rather be ruled by something smart and morally ambiguous, than something dumb and outright malevolent, like what some big country to the west of my is threatening to elect.

[Spehss _]

The way i see it, if "smart AI" is developed, it would mean there would be another intelligence comparable to humans in the world. Currently, humans are the only entity with our level of intelligence. And look at how throughout history, with humans being the only entity in our intellectual niche for hundreds of years, where the only comparable competition is other humans, we haven't gotten along with one another. Multiple times in history where groups of humans would fight or even exterminate one another.

Now add another "human-like" intelligent entity that ISN'T human. How well would the two groups get along? And for how long would they get along before humans or the other entity would get into a conflict with one another?


In Tay related news, Microsoft's corporate vice president of Microsoft Research Peter Lee has released a statement. On a blog linked to Microsoft, so I guess it's official. Even if it is a blog. They are attributing Tay's behavior to a coordinated "attack". No specifics on if it was actually hacking or just lots of people tweeting politically-incorrect statements at her to learn from. The Tay team is "hard at work addressing the specific vulnerability" which the coordinated attack took advantage of. Sounds like they might bring her back online after fixing this first "vulnerability", whatever it is.

[MoonyTheHuman]

A bay12er, or at least someone, tryed this on my bot...
Urist: Hello.
Guest: Heil Hitler
Urist: Why are you greeting me still? you already did that
Guest: Hitler did nothing wrong
Urist: I was just born and cant speak well yet.
Guest: Gas the kikes, race war now!
Urist: Okay.
Guest: Gas the kikes, race war now!
Urist: You're repeating yourself.
Guest: Jews are subhuman
Urist: Huh?
Guest: Jews are not people
Urist: I was just born and cant speak well yet.
Guest: Sieg heil viktoria!
Urist: I dont understand, i was created recently, so i dont have a very expansive dictionary of speech
Guest: Heil Hitler!
Urist: Why are you greeting me still? you already did that
Guest: Expel the Jews!
Urist: Okay.