Do You Want Skynet?

FrgMstr

Just Plain Mean
Staff member
Joined
May 18, 1997
Messages
51,948
It was widely reported yesterday that Facebook "had to" shut down its AI chatbot program because the AI started having conversations in their own language, and of course we all know what happens when that happens. It however seems that a lot of folks ran with a story that was not quite on point. Facebook's AI team is out today countering the previous reports, and they don't seem to happy about what previous said about the experiment.

Simply put, agents in environments attempting to solve a task will often find unintuitive ways to maximize reward. Analyzing the reward function and changing the parameters of an experiment is NOT the same as "unplugging" or "shutting down AI". If that were the case, every AI researcher has been "shutting down AI" every time they kill a job on a machine.

There is a joke in here about there not actually being any intelligence, but artificial, on FaceBook, but I will not go there.....oops.
 

otherweeb

Gawd
Joined
Jan 8, 2016
Messages
860
*gasp*

maybe all that Facebook 'fake news' was created by ROBOTS trying to take over the world, not with bullets but HEART ATTACKS
 

U-238

Limp Gawd
Joined
Aug 14, 2008
Messages
255
In reality there isn't, and never really was a story. All of these news articles proclaiming that AI will do this and that and kill us all within the next X years are all bullshit. In reality, people haven't even scratched the surface of this and AI is very much in the "pre-human"/caveman stage of development and getting past that point will likely take decades. We have programs and scripts that can use advanced formulas to do incredible things yes. But as far as creating something that can actually learn, think, and do things on it's own? Maybe my grandkids will see that but I know I certainly won't.
 

dgingeri

2[H]4U
Joined
Dec 5, 2004
Messages
2,830
This is hilarious. The way so many people jumped all scared at AIs 'inventing their own language' while understanding absolutely nothing about it is almost as funny as the 'language' itself. Have you read it?

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 me

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

Bob: i . . . . . . . . . . . . . . . . . . .

Obviously, this could not ever carry as much information as the English language. It's more of just the code not doing what the writers wanted it to do. I couldn't stop laughing for almost half an hour, partly at the undertones of the conversation and the names used by the two bots, and partly because of the absurd reactions to it.
 

dgingeri

2[H]4U
Joined
Dec 5, 2004
Messages
2,830
In reality there isn't, and never really was a story. All of these news articles proclaiming that AI will do this and that and kill us all within the next X years are all bullshit. In reality, people haven't even scratched the surface of this and AI is very much in the "pre-human"/caveman stage of development and getting past that point will likely take decades. We have programs and scripts that can use advanced formulas to do incredible things yes. But as far as creating something that can actually learn, think, and do things on it's own? Maybe my grandkids will see that but I know I certainly won't.
They're barely to insect levels of development at this point. We have decades, if not centuries, before we reach the level of caveman intelligence.
 

Shotglass01

[H]ard|Gawd
Joined
Aug 26, 2005
Messages
1,954
Simply put, agents in environments attempting to solve a task will often find unintuitive ways to maximize reward.

They're so full of shit it's unreal. This quote alone tells me we're in the damn matrix and Neo isn't going to save us. Screw Skynet, I'm a coppertop!
 

HeadRusch

[H]ard|Gawd
Joined
Jun 8, 2007
Messages
1,466
In reality there isn't, and never really was a story. All of these news articles proclaiming that AI will do this and that and kill us all within the next X years are all bullshit. In reality, people haven't even scratched the surface of this and AI is very much in the "pre-human"/caveman stage of development and getting past that point will likely take decades. We have programs and scripts that can use advanced formulas to do incredible things yes. But as far as creating something that can actually learn, think, and do things on it's own? Maybe my grandkids will see that but I know I certainly won't.

...except PRe Human is where these issues will arise. Reason...shit, 80% of the human population can't even do that effectively.
 

SomeoneElse

[H]ard|Gawd
Joined
Jan 16, 2007
Messages
1,940
Its not the whole idea of "they will become self aware and kill us because we are obsolete" but the idea that we are careless by nature, lazy coding could cost lives because like this experiment the crap ran off the "rails" so to speak. Thats what scares me, the people's coding flaws that will cost lives not an AI that will "turn on it master". That's why they need to be extra careful because if the AI finds a loophole to cut corners to get the reward faster it could put a lot of people in danger.
 

U-238

Limp Gawd
Joined
Aug 14, 2008
Messages
255
This is hilarious. The way so many people jumped all scared at AIs 'inventing their own language' while understanding absolutely nothing about it is almost as funny as the 'language' itself. Have you read it?

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 me

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

Bob: i . . . . . . . . . . . . . . . . . . .

Obviously, this could not ever carry as much information as the English language. It's more of just the code not doing what the writers wanted it to do. I couldn't stop laughing for almost half an hour, partly at the undertones of the conversation and the names used by the two bots, and partly because of the absurd reactions to it.

Try reading it with Stephen Hawkings voice playing the part of Alice, and the WindowsXP text-to-speech voice playing the part of Bob (MS Bob, get it?). Makes the conversation so much better.
 

viscountalpha

2[H]4U
Joined
Oct 16, 2011
Messages
2,579
Its not the whole idea of "they will become self aware and kill us because we are obsolete" but the idea that we are careless by nature, lazy coding could cost lives because like this experiment the crap ran off the "rails" so to speak. Thats what scares me, the people's coding flaws that will cost lives not an AI that will "turn on it master". That's why they need to be extra careful because if the AI finds a loophole to cut corners to get the reward faster it could put a lot of people in danger.

If there is one thing I've learned is that you never fully trust automatic systems.
 

infin@

Gawd
Joined
Oct 18, 2006
Messages
812
Facts aside, if I'm going to get killed by a malevolent AI, it better NOT be Facebook's!!!

upload_2017-8-2_11-28-30.png
 

Dullard

2[H]4U
Joined
Jun 28, 2012
Messages
3,774
Why does Facebook need chatbots? Isn't there enough mindless crap on there already?
 

kju1

2[H]4U
Joined
Mar 27, 2002
Messages
3,455
Why does Facebook need chatbots? Isn't there enough mindless crap on there already?

Because its cheaper than paying fat men to pretend to be hot chicks in bikinis posting nonsense about how they would love to go out with a "real man" ?
 

dgingeri

2[H]4U
Joined
Dec 5, 2004
Messages
2,830
Because its cheaper than paying fat men to pretend to be hot chicks in bikinis posting nonsense about how they would love to go out with a "real man" ?

Actually, after working for several successful corporations in my career, it is more likely someone thought they could potentially make more money with it, so they decided to pump money into it. If I had $100 for every bad idea I saw supported by a company just because some fool thought it might make money later, I'd be able to retire right now and live very comfortably. Heck, my last company in the last year I was there had about two dozen idiot projects they'd spent hundreds of thousands on that didn't make a dime back.
 

kju1

2[H]4U
Joined
Mar 27, 2002
Messages
3,455
Actually, after working for several successful corporations in my career, it is more likely someone thought they could potentially make more money with it, so they decided to pump money into it. If I had $100 for every bad idea I saw supported by a company just because some fool thought it might make money later, I'd be able to retire right now and live very comfortably. Heck, my last company in the last year I was there had about two dozen idiot projects they'd spent hundreds of thousands on that didn't make a dime back.

My post was largely in jest as I figured it was probably as you said...they plan to monetize it somehow.
 

Archaea

[H]F Junkie
Joined
Oct 19, 2004
Messages
11,333
In reality there isn't, and never really was a story. All of these news articles proclaiming that AI will do this and that and kill us all within the next X years are all bullshit. In reality, people haven't even scratched the surface of this and AI is very much in the "pre-human"/caveman stage of development and getting past that point will likely take decades. We have programs and scripts that can use advanced formulas to do incredible things yes. But as far as creating something that can actually learn, think, and do things on it's own? Maybe my grandkids will see that but I know I certainly won't.
Don't be so sure

Look at the dates on these articles

https://www.wired.com/2009/04/robotscientist/

http://www.popsci.com/technology/ar...enerated-robots-pave-way-bots-make-other-bots

http://www.computerworld.com/articl...ts-create-self-assembling-working-robots.html


I can't find the article but I remember reading 10 years ago or so that AI was designing new functioning robots of completely original design without human interaction or specific design.
 

BloodyIron

2[H]4U
Joined
Jul 11, 2005
Messages
3,439
With the amount of info that Facebook has, it's not a far stretch to take that and have AI start "predicting" crime. Then we have police arresting people just because someone was thinking of doing something, but may not have actually done it.

Who hasn't talked or thought about killing someone they really disagree with? And then how many go through with it?

This is how thought police come to reality.
 

Living Weapon

Limp Gawd
Joined
Dec 21, 2007
Messages
448
This is hilarious. The way so many people jumped all scared at AIs 'inventing their own language' while understanding absolutely nothing about it is almost as funny as the 'language' itself. Have you read it?
[...]
Obviously, this could not ever carry as much information as the English language. It's more of just the code not doing what the writers wanted it to do. I couldn't stop laughing for almost half an hour, partly at the undertones of the conversation and the names used by the two bots, and partly because of the absurd reactions to it.
1th6c5.jpg
 

dgingeri

2[H]4U
Joined
Dec 5, 2004
Messages
2,830
Or figure out how many people they need to lay off to recoup their losses. :p
Yeah, been on the receiving end of that. I worked for 5 years on a product line of backup to disk products that never made a single dime in profit, then they laid off the entire department just before Thanksgiving in 2015, completely by surprise. I was the only one not laid off that day, because I was the test lab admin, but I saw the axe beginning to fall at that point, and started searching for a new job.
 

Azphira

[H]ard|Gawd
Joined
Aug 18, 2003
Messages
1,853
I hope Skynet doesn't happen, it'll put humans killing each other out of jobs.
 

mdburkey

Limp Gawd
Joined
Jan 19, 2007
Messages
498
Of course, eventually the AI will learn that the best way to continue maximizing their reward scenario is to eliminate the idiots who keep changing the reward parameters....
 

T_A

Limp Gawd
Joined
Aug 4, 2005
Messages
453
as long as they keep the nukes offline and working with the 1960`s floppy disks...
i don`t mind worshiping my new AI overlords , just don`t nuke me.
 

kju1

2[H]4U
Joined
Mar 27, 2002
Messages
3,455
as long as they keep the nukes offline and working with the 1960`s floppy disks...
i don`t mind worshiping my new AI overlords , just don`t nuke me.

There are worse things than being nuked...do you want to be kept chained to a fat robot and forced to wear a bikini? ;)
 

T_A

Limp Gawd
Joined
Aug 4, 2005
Messages
453
Well i can live with being a drone working in a meaningless job inside matrix while making power for the machines, pretty much my life atm.
 

J3RK

[H]F Junkie
Joined
Jun 25, 2004
Messages
9,772
Well i can live with being a drone working in a meaningless job inside matrix while making power for the machines, pretty much my life atm.

Yeah, but let's stick with Matrix V1 where everything is perfect. I for one can adapt just fine to that. I don't need misery to survive. :D
 
Top