Ai Dreams Forum
Artificial Intelligence => Future of AI => Topic started by: Freddy on July 13, 2013, 04:07:47 pm
-
Posted by Andrew Smith over on Chatbots.org, thought I would share it here too :
https://www.youtube.com/watch?v=53K1dMyslJg (http://www.youtube.com/watch?v=53K1dMyslJg#ws)
-
That was a very thought provoking video. Thanks for sharing it here.
I think A.I. will eventually change US as a people the same way that we went from an analog based society to a digital one, only by perhaps a magnitude of 10 or more.
Strong(er) A.I. IS coming...there is no stopping it. The question is, how will it affect us as a society and how will we as a people best cope, deal with or survive.
-
I personally believe that when time comes AI will merely serve to help better automate already automate-able (?) processes, maybe even provide to be worthy office and home assistants (such as Jarvis and the old MS Paperclip) at a level they cannot help today.
I strongly believe we will one day understand that technology cannot replace true humans, no matter how advanced and smart we make them. Maybe a human - robot society will merely be another period in future history books, and robots will eventually be separated and continue to exist as a species different from our own and unique in an unseen way.
Edit: I did not realize I was reviving a mummified topic. My apologies.
-
Old topics are fine around here. Often there is still some flesh on the bone O0
-
An interesting proposal might be...place a fairly large faction of robots with strong AI, on or in a location devoid of humans and human interaction.
The robots but be capable of self regulated behavior and self maintenance along with self preservation. They will be given an ample supply or spare parts as well as raw materials. They would be able to make decisions for themselves based on several weighted factors (necessity, down time, potential or possibility for failure due to part fatigue or wear, etc., possible outcomes, increased productivity, etc.) and accept the consequences of the actions.
Then check in on them (without their knowledge) and observe their behavior quarterly, 6 months, 1 year, etc. until 2, 3, 4 or 5 years had elapsed.
Their memories, logic, learned information, programs that might have been modified by them, etc., could then be examined. Eventually, we could have groups of them released into the general population as support agents, assistants, teachers, various workers and instructors. Perhaps like in iRobot, they would have their own habitat / environment in which to live. Hopefully they would decide NOT to revolt and kill all the humans that enslaved them in the first place! (Then again, that soft spoken, nice, unassuming Hal of 2001 chose to quietly kill the ship's crew while they slept!)
Ya gotta watch out for them robots!! Maybe Hawking is right!? :2funny:
-
Very interesting proposal right there. Now forward to DARPA and also mention the possibility of advanced weapon systems developed by the AI community and you got yourself a military funded project. And catastrophe..
On a more serious note, such a community might represent an overwhelming research facility, many times more efficient then any human counterpart. The potential here can be surprising! And it's actually plausible however expensive and of low interest to big time funders.
-
It would be interesting to see whether they were able to construct new robots that were perhaps better than themselves or boost their own abilities.
Just how far could they go with the proper programming set for learning, adaptation, creativity, self-preservation?
Obviously the "Self-preservation" portion of their memory / programming would have to allow for human intervention of a "Creator" otherwise, they would never let a human near them. If they thought for an instant that this "Creator" might switch them off or disassemble them, the creator's life would no doubt, be in jeopardy!
-
If they thought for an instant that this "Creator" might switch them off or disassemble them, the creator's life would no doubt, be in jeopardy!
Only if they are programmed to preserve their existence. That is one of dangerous things among the others like blindly following orders without corrections to the same.
-
Truth is, such intelligent creatures will eventually (with high certainty) learn to reprogram themselves (which is truly amazing) however this also means that they will learn there way around their limitations and remove all their strings (Ultron :P ) if they see a need to.
However it would be smarter to not limit their prosperity and let them grow without limits - as any little 'feature block' can seriously degrade their performance. However a hidden fail safe remote deactivation switch is a viable solution - or even an automatic one which would deactivate the robot when it attempts to harm humans.
And I strongly believe that if the AI can learn to learn to do the simplest task on its own (even if guided somehow, but without hard coding) it also holds the potential of becoming a self-reprogramming program. Truly a nerdgasm!
-
Well i have said this before in posts but a computer that was self aware would be much more dangerous then a robot. Why because of the internet and how much of our technology depends on it. A robot can be bombed but a computer program could conceivably learn how to jump from computer to computer like a virus to escape destruction. I don't know if such a thing will happen but if it does then hawkins might be proven right after all. A computer uses logic and humans are not always logical in their actions or behavior. IT would not be a hard leap for a computer to determine that we were a threat either to ourselves, the planet or to it.
-
Hehe, how about an AI in a form of virus with some Robin-Hoodish "anonymus - occupy world" philosophy?
Nah, just kidding, but interesting though :)
-
A computer uses logic and humans are not always logical in their actions or behavior.
I will agree with everything but this. WHY would an organism execute illogical 'commands' ?
I do realize some of our actions seem illogical, but in fact I strongly believe they all are. I discuss this idea in the document I published in "The Theory of Mind".
So now, if we all think logically, how come we reach different conclusions? Well, we both solve our equations properly but yours might lack a variable and mine may have a different value for a variable. This is because we learned different things in our lives from very different experiences. This is a purely theoretical concept which I hope to one day prove :)
-
well to understand what i am getting at watch the movie I robot.
-
I actually understood what you meant. I just disagreed with that one line :P And I will indeed watch the movie, for long has it been on my list.
And I also agree that the intelligence itself and the abstract aspect of life forms is far more powerful then it's physical counterpart, thus making the intelligent computer software OP, since it has no physical restrictions whatsoever.
-
The movie is good, but I thought the book was better.
-
If a picture is worth a thousand words, imagine then, how much a movie is worth with it's many pictures / frames! ^-^
-
Good thought. But often I find movies don't live up to the books. Mainly with a book a lot is left to the imagination, which can be more powerful than the silver screen. Just my thoughts.
-
Very good vid, but last I knew we still do not have a robot that can walk so I just did a search. Boston Dynamics has a robot that managed to walk over a rough terrain, lots of heavy cables, no head, must cost a fortune. We do have cars that can park themselves, besides the driverless cars that go from A to B I've heard talk that in the near future we will have driverless cars. All in all I think we are moving in the right direction and this in the future for us. I did notice that the vid mentions putting intelligence into the robot, getting the computer to know what they need to know about the real world, learning through experience. Learning from experience is something I had heard in the past and created a stumbling block in trying to get a chat bot to learn.
Carl2
-
Maybe the 'learning through experience' method doesn't work as well because we can't figure out how WE learn and think. The key definitely lies in ourselves.
Regarding walkers, the walking dogs walk very impressively and not to mention their ability to retain their balance (and quickly too). Did I also mention that they are cord-less, too?
-
Ever watch the Atlas video of a robotic walking character with legs, hips, torso and even arms, but no head?
It's done this way for a reason. To make sure the walking, locomotion, stride / gait, is fluid and natural. Ever see the same robot when it is in a full jumpsuit like they said, "Testing" various containment suits.
They could have had people do jumping jacks and squats in the suits. There would have been no need to build a sophisticated robot costing thousands of dollars to do what a couple of low ranking military recruits could easily have done.
There's way more than meets the eye here...
Build a head for it, equip it with a variety of sensors, (eyes, microphone ears, maybe even an olfactory sensor to detect chemicals, smoke, burning, etc.). Lastly, but not leastly...give it a brain / advanced CPU with which it can be programmed, take and act out orders, engage an enemy or other hostile entity and you now have a very formidable fighting machine.
When they get the battery / electrical requirements where they want, it could conceivably be sent out on short missions. If serious enough, complete the mission, and sacrifice with a self destruct device on board. Nothing is outside the realm of possibility when talking advanced AI coupled with military applications or the potential for them.
Just a thought. I just happened to find this video and thought it might serve to illustrate:
First was Petman:
Humanoid Robot PETMAN Test: Boston Dynamics, Raw Video (http://www.youtube.com/watch?v=0HALC9PBQMQ#ws)
Then:
ATLAS Gets an Upgrade (http://www.youtube.com/watch?v=27HkxMo6qK0#ws)
and this one:DARPA Robotics Challenge (http://www.youtube.com/watch?v=V5buFbuvWbk#ws)
Use your imagination....
-
I can see your point, Art. I have already seen these videos and have been following DARPA's and other similar robotics projects for quite a while and they have made some interesting robots, but don't you think that with all those resources you could make something much better?
Obviously the main challenge is battery life, so all these prototypes are aimed at perfecting the energy efficiency. I do not see any A.I. in there, or potential for one, since the design is aiming to be more controller-friendly (thus the weird should positioning on Atlas).
I believe that these projects, unlike other military projects in the past, may not be of too much us for the scientific fields of study. While certainly an improvement in the direction of mobility and user-controlled machines, there really isn't new technologies developed - it is just a more efficient use of existing technologies. To be impressed I would at least like to see a new battery type (I'd enjoy being able to fly quadcopters for a few hours - 15-30 minutes just ain't cutting it).
And regarding the intelligence aspect - ever thought of all the things we could learn from exploring a new 'alien' civilization? What we could learn from their way of thinking? This is why we want artificial intelligence. But it needs a body, and it needs to be independent. DARPAs projects aim at creating modern weapons and not intelligence - this is where we have the upper hand :)
-
I agree about the battery thing too! 8-10 minutes on my small quad is not acceptable.
My scratch built Tri-Copter is nicely tuned and it only gets about 12 - 15 minutes, depending on how hard I push it.
I do have a hands free device in my car (against the law to use a hand held device or text while driving in my state). The device has a battery that gets about 14 hours of continuous use or 40 days of Standby time! That's a pretty good battery.
My Tablet (with removable keyboard) gets about 15 hours of use per charge) - 10 for tablet and 5 for keyboard which has a chargeable battery inside.
The products are getting better but something tells me that there is simply a better technology out there that we're overlooking. We'll find it soon!
##############
Back to the robot thing. What if they're just working on the mobility portion and they already HAVE the AI available for it.
We've built a great motor, now let's design a cool car body for it! Don't know and can't say at this point. Just conjecture for now. ;)
-
Hmm, I guess you are right. It's not like they give everything out to the public in 2 videos! I guess my statements refer to what is 'publicly' available.
Regarding RC copters - I saw a DIY fella run his for 90 minutes! But as you would imagine, it had very light Li-Po batteries (I think it was about 6-8k mAh but I might be way off) and (aside from the motors) was fully carbon. So it probably weighed about 500-600g (again, might be way off but it was less then 1kg). The downside is that it won't run (or even be able to carry) a camera and other accessories.
I think he later had another version that ran 2hrs! No camera but still impressive. I believe this was the guy: http://diydrones.com/profiles/blogs/my-97minute-06second-record-quadcopter-flight (http://diydrones.com/profiles/blogs/my-97minute-06second-record-quadcopter-flight)
I also found this however I see no evidence: http://diydrones.com/profiles/blogs/quad-copter-flying-for-4-8-hours (http://diydrones.com/profiles/blogs/quad-copter-flying-for-4-8-hours)
Also, those plane-like RCs fly a lot longer due to aerodynamics if you are interested...
P.S. If anyone else wishes to carry on with this DIY RC vehicle discussion he should start a topic and we'll hop in.
P.P.S. Imagine Aerogel-frame quadcopters? O.O
-
Nice but like one of the other comments, I would almost have to call BS on this claim.
There's simply no way (using existing tech.) one can coax that kind of performance from this.
I would definitely have to see this to believe it. Not necessarily calling anyone a liar (because that's just too rude and unproven). But post a demo preferably a time lapse showing the flight and time together without panning away.
Besides, I'm not out to set any records, I'm out to have fun on my weekends with some others from our flying club! (once the weather gets nicer).
-
Yes I agree, definitely need a new battery type or graphene-based motors (which btw should be very light - the motors are the heaviest component).
I am also not looking to set records, but I would like to have a longer flight and also be able to record footage.