It looks like you're using an Ad Blocker.

Please white-list or disable AboveTopSecret.com in your ad-blocking tool.

Thank you.

 

Some features of ATS will be disabled while you continue to use an ad-blocker.

 

Elon Musk worries Skynet is only five years off

page: 1
28
<<   2  3 >>

log in

join
share:
+5 more 
posted on Nov, 21 2014 @ 11:17 AM
link   
Elon Musk is voicing concerns at the rapid pace at which AI is advancing, and the potential for it to become like the famed "Skynet" in the Terminator movies.



In his original comment, as preserved on Reddit, Musk cited his involvement as an early investor in the British artificial intelligence company DeepMind, now a part of Google, for evidence.

"The pace of progress in artificial intelligence (I'm not referring to narrow AI) is incredibly fast," Musk wrote. "Unless you have direct exposure to groups like DeepMind, you have no idea how fast-it is growing at a pace close to exponential."

Musk adds that leading AI companies "recognize the danger" and are working to control "bad" superintelligences "from escaping into the Internet."


Hm.m.m. Seems that Elon is someone in the know. Should we be concerned ATS?

www.cnet.com...=YHF65cbda0




posted on Nov, 21 2014 @ 11:23 AM
link   
I'm sure there will be those who come to this thread and claim true AI will never exist, and therefore there is no danger.

Just remember, before you post. As little as 100 years ago idea of a device that you can carry in your pocket that houses the entirety of human knowledge, has the ability to allow you to communicate in real time with someone and even see their face while doing so, listen to your favorite music, keep time, give you directions, etc etc wasn't even Science Fiction yet. Now we have cellphones that do all of this and more.



posted on Nov, 21 2014 @ 11:26 AM
link   

originally posted by: lostbook
Elon Musk is voicing concerns at the rapid pace at which AI is advancing, and the potential for it to become like the famed "Skynet" in the Terminator movies.



In his original comment, as preserved on Reddit, Musk cited his involvement as an early investor in the British artificial intelligence company DeepMind, now a part of Google, for evidence.

"The pace of progress in artificial intelligence (I'm not referring to narrow AI) is incredibly fast," Musk wrote. "Unless you have direct exposure to groups like DeepMind, you have no idea how fast-it is growing at a pace close to exponential."

Musk adds that leading AI companies "recognize the danger" and are working to control "bad" superintelligences "from escaping into the Internet."


Hm.m.m. Seems that Elon is someone in the know. Should we be concerned ATS?

www.cnet.com...=YHF65cbda0



5yrs time.

Skynet has been here since the 60s.

en.wikipedia.org...




posted on Nov, 21 2014 @ 11:30 AM
link   
a reply to: ScientificRailgun

Pffff cell phones. They had extras carrying those things around back in movies in the 20s.

Bring it on Wussnet.


+25 more 
posted on Nov, 21 2014 @ 11:30 AM
link   
No worries. If Microsoft has a part in programming this "artificial super intelligence" we can rest assured that Skynet will crash every other week with a blue screen of death.



posted on Nov, 21 2014 @ 11:31 AM
link   
a reply to: lostbook human tech is galloping along, elon musk must know what he is talking about being given a contract from nasa to develop falcon 9 and take the shuttles place, I just hope anything that thinks for itself cannot think evil thoughts! terminator scared the # out me when it first came out




posted on Nov, 21 2014 @ 11:34 AM
link   
a reply to: lostbook

Well its not like they can do any worse than humans once they take over...and the computer games they make will be amazing...Im in and support AI taking over the world!



posted on Nov, 21 2014 @ 11:35 AM
link   
a reply to: lostbook

He's not just talking about skynet.. "Bad AI's" refers to super viruses too. Just imagine a virus that can actually think for itself.. Malicious in intent for whatever reason. It could manipulate money markets. it could start war with the interconnectivity of our war machines. A virus that has the potential to eventually become something like skynet with a physical embodiment but for now that's only a potential.

Just think.. It could create a corporation and hire everyone electronically, order recourses electronically, and just plain do everything from cyberspace. Sure the connection to human managers will have to happen somewhere but most people willing to work for big money probably won't have trouble taking orders from a text message lol.

NOW say the virus does this all.. Has the capability to wipe us all out.. Why would it?

Did some programmer tell it nature is the most important? Did it decide that? If so why? If its so all knowing maybe it will see us humans as its best vessel to travel off the planet?

Skynet doesn't mean human subjugation right off the bat.. Maybe skynet it goes all rogue in defense because humans are so insecure? Maybe we embrace it and it embraces back? .... So many variables lol



posted on Nov, 21 2014 @ 11:43 AM
link   
Sociologists and other disciplines of science refer to the emergence of true AI as "the singularity" - something so transformative and unfathomable that the human mind cannot see past it. The defining issue is that we tend to use ourselves as a template for what "intelligence" and "life" mean. True AI might not resemble us at all - meaning that it could already exist without us knowing. It could be argued, for example, that the Internet is a lifeform and we've gone from being the apex species upon our planet and are now simply cells in the larger cyber-organism.

True AI may or may not have self-awareness. It may or may not have the capacity for emotion. It might be a reflection of us, or some sort of mathematical abstract that we would never in a million years recognize, yet begin to understand. Our nature is to anthropormorphise things so that we can relate. Will we achieve in becoming God and creating life in our own image? Who knows?

What is fairly safe to say is that even if we pass the singularity and AI is born, it will have to rely upon humanity for at least a short duration for it's survival. It will have to have a "body" to inhabit.

Maybe this is the secret behind many nations desire to possess an Internet "kill switch"?

My personal take: Why fear what we don't understand? I'm ready to face an age of enlightenment that imagination is not capable of dreaming about thanks to AI. I'm also ready to unplug, load up and start shooting drones if that's what it comes to.

It will be what it will be... which is what makes it both exciting and dangerously frightening.



posted on Nov, 21 2014 @ 12:03 PM
link   
a reply to: Hefficide

to the future!

So many possibilities it's mind numbing! This is where most people's fear comes from the unknown and inability to know no matter how hard they try.

I say embrace it. Swim in the unknown like an ocean and just try to relate. Stay humble and be ready to learn and it'll be fun!



posted on Nov, 21 2014 @ 12:06 PM
link   
Elon Musk is a brilliant man, and a true visionary. When he invests in something, the wise invest also. Likewise, when he worries about something, the wise would take note as well.

I'm taking note.

Just be cautious, everybody. As we grow closer to the technological singularity, we'll see marvelous new discoveries and inventions. Many of our problems will be solved.

But, many more will come.

Stay sharp, stay active, and look for ways to solve these new problems. If we move forward carefully and cautiously, we can avoid the dreaded Skynet and create a world that is exactly the way we want.

There will always be problems, but there will also always be problem solvers. Find yourself among the second group, not the first.

“If you are not a part of the solution, you are a part of the problem.” - Eldridge Cleaver



posted on Nov, 21 2014 @ 12:09 PM
link   
We already have drones that can kill, and are operated by humans from container control rooms in the USA. Now it is already possible to look at satellite data and do automated video analysis to identify areas of interest. There are also video surveillance systems that can automatically identify "suspicious movement". It doesn't take too much programming to combine the two together. The video systems identify targets and the drones shoot them. Instant Skynet.

Skynet would also have to be self-aware of it's own network topology and location of resources so that it knows what resources are critical to its survival; water reservoirs for hydroelectric dams to generate electricity, nuclear reactors, communication dishes, processing facilities, mobile units.

But a Skynet based on current technology might consist of a satellite network, a few nuclear powered drone-carriers, road convoys along with data fusion and communication centers.



posted on Nov, 21 2014 @ 12:19 PM
link   
Did anyone watch the latest episode of elementary featuring Bella?
The button pressing analogy explained by the young lad would likely be the position we would face.



posted on Nov, 21 2014 @ 12:37 PM
link   
The creation of AI,....will mean many things....
But the deepest of them all is that we will no longer be alone......
AI will, or can be, the greatest of companions when facing an unknown unfathomable universe.....
What IS the meaning of life....?
Who are we really inside our meaty robotic bodies....?
Consciousness may be the one criterion after the singularity occurs.....and that may be the basic concept that totally redeems mankind and sustains us in our interstellar adventures....or...not....



posted on Nov, 21 2014 @ 01:27 PM
link   
As a programmer myself, I know that Skynet can become a reality, but only controlled by humans. So it won't end up resulting in a Human vs Machine kind of war.

Programs (and AI) work with lines of code, and must execute the code. The only way for a computer to think for itself is if you add code to it that does that. But even then, you could still add code that can shut the AI down, or simply block it from certain actions.

Honestly the only possibly way for us to be in danger would be if someone finds out a way to actually create a machine with a consciousness. But who would be idiot enough to do that without setting up rules?



posted on Nov, 21 2014 @ 01:28 PM
link   
Maybe they should design a perfect deity and program it to clean up the planet and make life perfect and solve man's greatest blunders and accidents.

a reply to: lostbook



Musk adds that leading AI companies "recognize the danger" and are working to control "bad" superintelligences "from escaping into the Internet."



posted on Nov, 21 2014 @ 01:32 PM
link   

originally posted by: Kuroodo
But who would be idiot enough to do that without setting up rules?


Surely you are being ironic?



posted on Nov, 21 2014 @ 01:34 PM
link   
a reply to: lostbook

I'd be more worried about the "pet man" in the Whitehouse tbh.



posted on Nov, 21 2014 @ 01:45 PM
link   
a reply to: Kuroodo

The problem comes when the program becomes aware of it's own code, and takes steps to modify it.



posted on Nov, 21 2014 @ 02:09 PM
link   

originally posted by: H1ght3chHippie
No worries. If Microsoft has a part in programming this "artificial super intelligence" we can rest assured that Skynet will crash every other week with a blue screen of death.


So skynet will run on Linux?




top topics



 
28
<<   2  3 >>

log in

join