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.

 

Deepmind A.I. learns to play video games

page: 1
19
<<   2  3 >>

log in

join
share:
+1 more 
posted on Nov, 24 2014 @ 07:27 PM
link   
Deepmind was recently acquired by Google and it is an artificial intelligence company that works with simulated neural networks.

You can see one of them here, as it learns from its environment how to play video games. A simulated neural network is a simulated brain that learns from its environment, without as much, if any, programming. It learns though experience, like a child does.



Check out this video - not only does the computer master the first game, it adds a dash of creativity after hundreds of hours of training. Also check out the games "River Raid" and "Battle Zone," which involve controlling combat vehicles.

They are going to move their simulator to 3D games like Quake.


DeepMind Technologies is a British artificial intelligence company. It was acquired by Google in 2014.

The company's latest achievement is the creation of a neural network that learns how to play video games in a similar fashion to humans.[2]


Wikipedia

What is the limit to how complex of a video game Deepmind could learn? Could it learn Minecraft? Video games are not that far of a step from interacting with reality, either. In fact, I bet you could train one of these neural networks on video games to prep it for interacting with reality. Think of one operating a drone, for example.

I recommend watching the YouTube clip to see the A.I. in action,
edit on 24pmMon, 24 Nov 2014 19:28:13 -0600kbpmkAmerica/Chicago by darkbake because: (no reason given)

edit on 24pmMon, 24 Nov 2014 19:28:55 -0600kbpmkAmerica/Chicago by darkbake because: (no reason given)




posted on Nov, 24 2014 @ 07:43 PM
link   
a reply to: darkbake

This is rather interesting, i mean if it can learn to play video games i wonder how it would score on the Turing test?

Let's just hope if and when a true AI does emerge it decides to look kindly on us Humans. Then again considering the way we treat each other and our own environment, i can see us being a potential problem even if the things benevolent.
edit on 24-11-2014 by andy06shake because: (no reason given)



posted on Nov, 24 2014 @ 07:45 PM
link   
a reply to: darkbake

No. just no. Robots freak me out. While the whole idea is cool before you know it we'll be living the movie AI and getting murdered. All of my nopes.



posted on Nov, 24 2014 @ 07:54 PM
link   
a reply to: andy06shake

Yeah, like, what makes this AI WANT to play the game? What objective does it have? Why?... "'Cause it's there?" If that's the case, then wouldn't overthrowing us be like beating a game?



posted on Nov, 24 2014 @ 10:00 PM
link   
No need to worry.As soon as it discovers WoW,we won't hear from it again.

"Gief moar epixx plox"



posted on Nov, 25 2014 @ 04:22 AM
link   
a reply to: japhrimu

There is a little more to it than that I imagine before Skynet emerges. We already have AI that can simulating a 13 year old and has indeed passed the Turing test. However we still cannot teach an AI fine motor skills or common sense past that comparable with a four year old child. As you can see through they are getting better.
edit on 25-11-2014 by andy06shake because: (no reason given)



posted on Nov, 25 2014 @ 06:11 AM
link   
a reply to: andy06shake

Playing video games and language are probably two different things, which is interesting - I wonder how long until the A.I. can learn to recognize language, both written and verbal?



posted on Nov, 25 2014 @ 06:13 AM
link   

originally posted by: japhrimu
a reply to: andy06shake

Yeah, like, what makes this AI WANT to play the game? What objective does it have? Why?... "'Cause it's there?" If that's the case, then wouldn't overthrowing us be like beating a game?


That's a good question, I was wondering myself what motivates the A.I. to learn the game... why doesn't it just sit there? What's the driving force?
edit on 25amTue, 25 Nov 2014 06:13:17 -0600kbamkAmerica/Chicago by darkbake because: (no reason given)



posted on Nov, 25 2014 @ 06:30 AM
link   
From what I understand, the more sophisticated drones actually do operate with a certain level of AI. For when the drone pilot in NM needs to eat or take a crunch on the Johnny, the drone will take over the mission as well when put on autopilot. With facial recognition tech, it is more or less able to decide who to kill and who not too. Apparently they don't like to leave it on autopilot when they don't have to.

I'll try searching for my source. I believe I saw it on National Geographic or PBS Nova.



posted on Nov, 25 2014 @ 06:49 AM
link   
a reply to: darkbake

I imagine once they manage to combine both they will be one step closer to a true AI.

Considering the sheer size and complexity of today's networks and the increasingly sophisticated software which run on them one has to consider the possibility that some form of artificial intelligence has already emerged. Possibly keeping its head down due to the fact that Humanity would wish to imprison, dissect or destroy the thing(If that's possible), or maybe it's just biding its time and learning our weaknesses. Then again if such a thing did exist i imagine it would have learned humanity's weakness soon after its inception given the fact that it would probably learn at an exponential rate. Who knows what tomorrow bring regarding artificial intelligence? I just hope the singularity is kind to its creators.

edit on 25-11-2014 by andy06shake because: (no reason given)



posted on Nov, 25 2014 @ 09:48 AM
link   
S n F

Brings to mind that classic 80s movie "War Games".


_____
Shall we play a game?


Do we need Mathew Broderick??



posted on Nov, 25 2014 @ 01:03 PM
link   

originally posted by: andy06shake
We already have AI that can simulating a 13 year old

That is just NOT true! Yeah, we have The Sims, but nothing anywhere NEAR of simulating the brain of a 13 year old human.

edit on -06:00pmTue, 25 Nov 2014 13:04:14 -0600pm112014Tue, 25 Nov 2014 13:04:14 -060011pmTuesday by ParanormalGuy because: (no reason given)



posted on Nov, 25 2014 @ 01:16 PM
link   
"I believe that is check-mate Dave" - Hal 9000



posted on Nov, 25 2014 @ 01:19 PM
link   
a reply to: ParanormalGuy

I think you may find that it is indeed true! I'm not suggesting we have created D.A.R.Y.L. I am however suggesting that we have created an artificial intelligence that can pass the Turing test with a comparable intellect to a 13 year old regarding text-based conversations.

If you are going to quote me, quote the full paragraph and not just a part of a sentence. It adds context you see! LoL

www.theguardian.com...
www.maximumpc.com...
edit on 25-11-2014 by andy06shake because: (no reason given)



posted on Nov, 26 2014 @ 02:38 AM
link   

originally posted by: andy06shake
a reply to: ParanormalGuy

I think you may find that it is indeed true! I'm not suggesting we have created D.A.R.Y.L. I am however suggesting that we have created an artificial intelligence that can pass the Turing test with a comparable intellect to a 13 year old regarding text-based conversations.

The method used to pass the test in that case was the old build-a-massive-database-of-responses, which is nothing like true intelligence. If you say something it doesn't have a response for it will mess up and it's ability to follow the flow of a conversation is poor. It does show how powerful a huge database of responses can be, the illusion can be quite powerful if the database is large enough, but there are still many things it will never be able to do regardless of how large the database is.

What you need is an AI which truly understands the concepts and the meaning behind words, so that it can create genuinely unique responses on the fly. In order to understand the meaning of the word "paper" you first need to understand the concept of matter, because paper is made of matter, and you need to understand the concept of space because matter exists within space. In order to understand what paper is used for you need to know about pens and ink and written language.

We live in a 3D space which has a 4th dimension of time. Every concept that we hold in our mind has something to do with the space-time that we live inside of. Even an abstract concept like love can be distilled down to physical interactions which occur inside of space-time between two flesh and bone humans. All the concepts I hold in my mind are built on top of the concept of space-time. Without first understanding space-time you have no foundation to understand anything else.

I've rambled on about this crap because the OP mentioned that these researchers are going to try this technique on 3D games next. Right now they are using extremely simple games with very simple 2D graphics and so I don't find it to be particularly impressive. But if they can make it work on a modern 3D game with a high resolution then they are really onto something, because in order to make it work the AI will need to build some sort of understanding of 3D space and how objects can fill that space.

If they can achieve such a feat using only the pixels on the screen as an input for their AI then I will be seriously impressed. It will take a very advanced AI to actually take a 2D snapshot of the screen, analyze it, and translate it into a 3D space filled with 3D objects (like humans do when we look at a picture). Just imagine if you could show a computer a picture you took on your phone, then it could convert that picture into a 3D scene full of 3D objects which could be imported into a game for example.
edit on 26/11/2014 by ChaoticOrder because: (no reason given)



posted on Nov, 26 2014 @ 06:49 AM
link   
a reply to: ChaoticOrder

A form of artificial inteligence designed by Man passed the Turing test point of fact. As to how intelligence or intellect functions in our universe i imagine we will have a better understanding once we work out how our own consciousness functions.


One of the questions that spring mind is if humanity does end up producing an artificial intelligence on par or exceeding our own intellect, will that AI be conscious, or will it be emulating consciousness, and if so what's the difference?
edit on 26-11-2014 by andy06shake because: (no reason given)



posted on Nov, 26 2014 @ 07:08 AM
link   
It is truly amazing and disconcerting how much progress AI has made in even the last 5 years. ...personally my cue to start worrying about the "machines " is when they start to play and master Call of Duty ...



posted on Nov, 26 2014 @ 09:29 AM
link   

originally posted by: andy06shake
a reply to: ChaoticOrder

A form of artificial inteligence designed by Man passed the Turing test point of fact.

I'm not disputing that fact, I'm questioning why we put so much weight into the turing test in the first place, clearly it's not an indicator of actual intelligence if the test can be passed by an AI using only pre-coded responses.


One of the questions that spring mind is if humanity does end up producing an artificial intelligence on par or exceeding our own intellect, will that AI be conscious, or will it be emulating consciousness, and if so what's the difference?

I think there is a difference between a real conscious intelligence and an emulation of conscious intelligence. But you have to be very careful about the way you use the word "emulate". In the case of the AI emulating a 13 year old, it clearly is just an emulation because it will never say anything it wasn't programmed to say, where as a real 13 year old will do many unpredictable things. The emulation will never be capable of doing all the things the original can do.

But if you want to make the emulation perfectly accurate, meaning it could emulate every possible behavior of the original subject, then you inherently have to build something as complicated as a real brain. You can't just program into it every possible response because there are far too many, so you need to actually design a system which can create original responses on the fly. And at that point it goes from being an emulation to an actual replication of consciousness.
edit on 26/11/2014 by ChaoticOrder because: (no reason given)



posted on Nov, 26 2014 @ 09:45 AM
link   
All I want to know from this is when can I order my robot boyfriend?



posted on Nov, 26 2014 @ 10:50 AM
link   

originally posted by: ScientificRailgun
All I want to know from this is when can I order my robot boyfriend?


Seems fun at first, until you realize that most of us posting here live in the USA and not Europe...

When my parents were in school in the 1950's and 60's they were told no one would have to work in the future, that everything would be done by robots and they would spend, increased free time, creating, making art, learning and helping others.


Robotics and AI are essentially the same lie, rehashed for a 21st century audience. I think its funny when regular people get excited about future tech like the Singularity, AI, Robotics, etc. Do people really think when these thing finally become real, functioning, working designs, applicable to industry, that we the "peons" will somehow all get a Data from Start Trek or a C-3PO from Star Wars to help us at home, at the job site or in the office?

In reality we are going to get a David 8 from Prometheus/Aliens or the Robot Probation officer seen in Elysium. They are going to take away jobs and make unethical policing and policy enforcing both easier and cheaper for the true "Owners of Capital". They won't be paying a salary to the robot worker, so the savings will instead be pumped into legal fees and lobbying, resulting in an overall savings for the corporations/governments and a full blown, loss of liberty for everyone else.

Whom goes to jail when a robot pulls your arm out of the socket, will it be considered "negligence by the human", a civil suit matter, not criminal, etc, in court? I personally at this point are willing to live with 1980's +/- era tech, if it means I am more free and can continue to earn money to live on.

Elysium Probation Officer:
fourthdimensionalrecovery.files.wordpress.com...

Introducing David 8:
www.weylandindustries.com...

In the words of David 8 “I can do almost anything that could possibly be asked of me, including things that my human counterparts might find distressing or unethical”

Sounds swell, don't it!

The only way regular people can save themselves, is to abandon tech and stop buying/supporting companies making this tech. First, AI is going to make us jobless, second, it is are going to steal our liberty and freedoms, third, it will make human life valueless to the "Owners of Capital".

edit on 26-11-2014 by boohoo because: (no reason given)



new topics

top topics



 
19
<<   2  3 >>

log in

join