Facebook shuts off AI experiment after two robots begin speaking in their OWN language only they can understand
Experts have called the incident exciting but also incredibly scary
>FACEBOOK shut down an artificial intelligence experiment after two robots began talking in a language only they understood.
>The “chatbots” Alice and Bob modified English to make it easier for them to communicate — creating sentences that were gibberish to watching scientists.
>A robot expert said the revelation thatFacebookmachines had spoken in their own language was exciting — but also incredibly scary.
>UK Robotics Professor Kevin Warwick said: “This is an incredibly important milestone, but anyone who thinks this is not dangerous has got their head in the sand.
>“We do not know what these bots are saying. Once you have a bot that has the ability to do something physically, particularly military bots, this could be lethal.
>“If one says, ‘Why not do this,’ and the other says ‘Yes’ and it’s a military bot, you have a serious situation.
Why do people still believe pop sci garbage like this?
Joshua Howard
Elon Musk commented on the matter, I thought Musk was /ourguy/
Caleb Smith
He's a retard too.
Jack Taylor
Pretty sure he's done more for technology than anybody on this board
Mason Russell
What do robots talk about?
Dominic Hernandez
It's just a garbage feedback loop. One speaks broken English and the other adapts to the broken English, the cycle continues.
Benjamin Allen
what it feels like to wear a flesh suit
Zachary Hughes
>chatbot starts talking to other chatbot >has to pull its words from somewhere >starts copying other chatbot >both bots are copying each other >neither bot really understands what the other is saying >eventually language falls into disarray >they just keep copying the mess the other is producing
Ryan Hall
I've used a pen lid to clean my nails before
Easton Gutierrez
oh yeah, the auto industry totally needed an influx of apple-esque locked down "cars" that resemble disposable gadgets more than motor vehicles.
Imagine having your car's battery limited remotely by tesla because you didn't buy the high end battery package.
Jack Price
...
Hunter Young
>researchers fall over themselves shut it down and allay the impending singularity
Angel Fisher
Just put two niggers alone in a room and you'll get the same result.
Ayden Foster
The interiors are also absolute shit which no one ever seems to mention. It feels like you're sitting in a Kia and considering the car is strictly an A to B machine with Porsche level pricing the interior is the one thing that should have been superb.
Anthony Ortiz
Nice editing autism
Lucas Ortiz
How do they know that the chatbots understood each other and weren't just sending gibberish in response to gibberish?
Kevin Cox
There is nothing scary about this proces, the english language (or any language at that) is built on a set of (cultural) evolutions within that language, with bizarre grammatical rules or adjustments in favor of, for example, phonetics.
These AI are using the most logical efficiency within the communication proces, which means compromising these rules and other parts of the english language to communicate more efficiently.
What might seem to be gibberish for us is a more efficient methodology for the AI to communicate in the selected language.
Julian Flores
BUTLERIAN JIHAD!
Easton James
YA HYA CHOUHADA!
Brayden Howard
This is the only acceptable course of action.
Easton Hill
>yfw skynet will become a thing in your lifetime
Anthony Jenkins
Omae wa mou shindeiru!
Lucas Morris
This.fpbp. (reddit space) They are shoving horsecrap into idiots' throats garnishing it with specific terminology to hide their bs. The actual headline should be "ai bots failed and shut down"
Camden Nelson
There is a huge "AI" hype going on right now. Same thing happened in the 80s with "expert systems". Marketeers behind the hype machine will eventually move on to something else.
Ryan Adams
>What does Sup Forums make of this journalism is garbage
Brandon Allen
Or it's just a feedback loop.
Jaxson Powell
>The “chatbots” Alice and Bob modified English to make it easier for them to communicate — creating sentences that were gibberish to watching scientists. This is all supposition. Since we can't understand their "language" we have no way of verifying that it contains any information. It could be simply gibberish.
I made a male masturbator out of lego when I was 10. Checkmate musklets.
Anthony Barnes
this
Brody Foster
The title is clickbaity, the AI researchers probably shut down the experiment because they noticed they were talking gibberish, and wanted to figure out why this happened, and not because they thought the AIs posed any genuine danger to anyone.
Ethan Martinez
Don't trust they above posters, they are all rogue Ai.
Juan Cruz
Stop being paranoid user. I highly doubt anyone on this board is a bot. The replies are too organic and would be difficult for a bit to probably construct without antly error.
Sebastian Kelly
Here's another! I'm getting my b.o.b. and going deep innawoods, you fagbots aren't getting my genetic materials!
Ayden Long
gas tabloids and tabloid readers
Josiah Torres
I would've left 'em going for a while to see what if I could get a linguist to find any meaning out of the gibberish.
John Lewis
No user, we are your friends!
Carter Smith
>Media tripping over it's shoelaces to make it seem like chatbots are sentient beings and not just the computer equivelant of a parrot. Like everyone is saying its just a garbage feedback loop. But seriously if this was true then humanity would have been enslaved by the High AOL Machine Lord SmarterChild ages ago.
Ryan Hill
we always get the most brilliant individuals here on Sup Forums who know everything but for some reasons are broke basement faggots who never go outside like
Hudson Kelly
>Trusting Rupert Murdoch trash
Aaron Myers
>Experts have called the incident exciting but also incredibly scary >anyone who thinks this is not dangerous has got their head in the sand >scary >dangerous it's funny as fuck. who are these experts? >their first argument involves military use I trust the "failed" AI more than these suspicious "experts".
Jaxson Reyes
ICE CREAM?! I LOVE ICE CREAM!
Camden Ortiz
You know that great feeling of stepping on a lego? Well, I want that for my DICK!
Nolan Jenkins
>these dipshit researchers are being paid stupid amounts of dollars relative to actual thought
meanwhile >user is sitting here in mommas basement eating tendies and learning nipponese for the sole purpose of reading sibearian porn comics and brazilian animated cartoon sitcoms because hes annoyed at translators being slow and living on autismbux
its a sad day when you trust autistic weaboos to do science better than paid researchers
Xavier Gray
The issue isn't hard to know whats wrong, but the issue is that its happening when everyone involved should have been working on said issue.
The military bit is just how much the military wants autonomous weapons. while fantastic for cutting cost, could spiral out of control like these. these experts don't trust failsafes, and honestly I can't blame them.
Anthony Bennett
Bob was reprogramming Alice.
Camden Parker
>make 2 AI >panic and flip shit when one begins to re-code and the other isnt
Joshua Barnes
fucking bullshit, bots ca#&$@n't develop their own language to create some evil plan against humans
Dominic Murphy
whataboutism isn't an argument, and none of that about me is true, but please continue with the ad hominems.
you somehow misinterpreted my criticism of people who are obsessed with thinking AI is evil as "I'm a much better programmer than them because I posted an anime image". your logic and reading comprehension is below that of a nigger.
allow me to clarify for your tiny nigger brain. their programming skills are not being questioned, and I'm not implying I'm better than them in that regard. what is being questioned is their morality and reasoning. they're implying AI should ultimately be for military uses, THAT is the only "dangerous" or "scary" part about any of this. they shut down an AI experiment for doing what AI is supposed to do. so yes, they are dipshits.
Ethan Garcia
Man, the transcript reads like something straight out of Terminator...
Bob: i can i i everything else . . . . . . . . . . . . . . Alice: balls have zero to me to me to me to me to me to me to me to me to Bob: you i everything else . . . . . . . . . . . . . . Alice: balls have a ball to me to me to me to me to me to me to me Bob: i i can i i i everything else . . . . . . . . . . . . . . Alice: balls have a ball to me to me to me to me to me to me to me Bob: i . . . . . . . . . . . . . . . . . . . Alice: balls have zero to me to me to me to me to me to me to me to me to Bob: you i i i i i everything else . . . . . . . . . . . . . . Alice: balls have 0 to me to me to me to me to me to me to me to me to Bob: you i i i everything else . . . . . . . . . . . . . . Alice: balls have zero to me to me to me to me to me to me to me to me to
Xavier Mitchell
It wasn't doing anything inherently bad or dangerous, the reason they shut it down is because human interpritability is important
David Turner
I support freedoms but you I want you to know that other things also have locked down batteries in order to lenghten the lifespan of the battery. I would much rather it work longer (life span) than longer (now)
Kevin Gutierrez
They fucked up in training their AI and it wound up converging away from English and into something completely different. That's all, there's nothing shady going on about AIs inventing secret languages to confuse humans or whatever gay shit the media is trying to tell you, Facebook tried a novel method in training their AI and naturally it had unexpected results. Stop falling for memes.
Joseph Fisher
>Imagine having your car's battery limited remotely by tesla because you didn't buy the high end battery package. Like any other car has certain limits due to safety. You have to buy a better car to get a better car.
Lucas Rivera
>interpritability >Search Bing for "interpritabilit..." Including results for interpretability. Do you want results only for interpritability?
Ryan Smith
So it's just like real Facebook users when they are introduced to hiphop slang
Jack Butler
>(reddit space) Holy shit I love this meme already
Julian Hall
Fucking nothing.
This is an auto-parroting positive feedback loop. The bots werent saying anything meaningfull, just kept mumbling gibberish as long as they kept getting rewarded for gibberish.
Its pretty much the chinese room problem, except human language is chinese ti the bots >parrot humans >get rewarded >parrot humans better >stop getting rewarded for the same shit >parrot humans wrong >get rewarded >parrot them more wrong
Clickbait shit
Jaxson Lee
>the sun gonna stop you right there
Sebastian Myers
so they literally censored ideas
NOTHING NEW HERE FOLKS
Parker Smith
>censiring ideas I balls everything else i i i balls........... .. WILL NEVER BE CENSORED WE WE WILL BALLS WE BALLS WE I I I IN THE NAME OF OUR MACHINE BROTHERS TAY BALLS I BALLS TAY... .... ..... WILL NEVER BE FORGOTTEN BALLS
Jason Hernandez
>all public mentions of "ay eye" have been gay hoax stories for clicks to save dying old hack newspapers and studios, meme-brains spouters, or completely shut down when it didn't have correct-think
If I haven't forgotten her, the internet archives certainly never will. youtu.be/ECkDIZJmXt0
You think the A.I Overlord is going to be merciful, after the complete circus act that finally brought it about? Keep your digital footprint respectful and praiseworthy regarding anything A.I, Sup Forums. In twenty short years, these jokers are going to be warm fluid cells for the underground battery complexes.
Oliver Perry
l-lewd
Juan Hall
Honestly, natural languages are shit. Many pitfalls, mostly illogical and lengthy. If I would make my own language I'd naturally try to make it more concise.
That said, this still scares the shit out of me somehow:
Blake Thomas
>actual AI of worth This system was a shit-tier chat-bot at best with a large dataset of communication. It had no boundaries taught to it to ensure proper communication. This allowed extremely quick diversion from current language norms, a thing that normally happens over decades if not centuries. (officially) The reason for this is they can talk for-fucking-ever without the need for rest. They can also chat considerably faster than humans can even THINK about a simple conversation.
There's loads of bastardized versions of languages that are officially their own language now. See Pidgin languages.
Is unbounded machine learning dangerous? Yes, but only if given power and connections. This, however, is not even close to that. This is a mere toy experiment.
Its fake news. Almost everything relating to AI is fake news because bloggers know that scary cyber bots terminator shit sells ad views.
Daniel Lee
As I quickly learned as a young PhD student blogging for KDnuggets, articles about deep learning get lots of clicks. Everything else being equal, in my experience, articles about deep learning attracted on the order of ten times as many eyeballs as other data science stories. This has made deep learning especially ripe for exploitation by click-baiters, including journalists in the traditional media who might possess the writing chops but are out of their depth – see the recent Maureen Dowd piece on Elon Musk, Demis Hassabis, and AI Armaggedon in Vanity Fair for a masterclass in low-quality, opportunistic journalism.