?

Log in

Android videos - The Singularity

> Recent Entries
> Archive
> Friends
> Profile

October 2nd, 2006


Previous Entry Share Next Entry
mathemajician
02:25 pm - Android videos
Every year the robotics videos I see just get more and more impressive. These are via spoonless.







With robotics developing so fast, and CPU power and memory technology still growing exponentially, the only thing that might stop androids taking to the streets 10 years from now is that we still don't know the algorithms required to give them enough intelligence.

(6 comments | Leave a comment)

Comments:


From:dirkcjelli
Date:October 2nd, 2006 12:32 pm (UTC)
(Link)
the only thing that might stop androids taking to the streets 10 years from now is that we still don't know the algorithms required to give them enough intelligence.

I think that is quite a significant holdup... but ten years is a long time in terms of science/research/etc.
[User Picture]
From:mathemajician
Date:October 2nd, 2006 12:46 pm (UTC)
(Link)
Time will tell.

I think one interesting possibility is that nobody figures out the intelligence side of things for the next 15 years, during which time amazing robotics, CPU and memory technology is developed. In other words, all the hardware is there, it's just missing the algorithm. People see a few more robots around in place but not all that much more...

...but then there is an AI breakthrough and somebody comes up with a vastly more powerful algorithm. Given that the hardware is so powerful, the new algorithm would be deployed extremely rapidly. I imagine that such a quick transition could cause a large social upheaval.
[User Picture]
From:baikonur
Date:October 2nd, 2006 02:30 pm (UTC)
(Link)
The hardware is powerful, but that isn't the real holdup. Even given the existence of a real AI, there are several barriers to real word deployment.

The first is cost. Most of these robotics developments are either academic projects or the pet project of major corporations, usually at the cost of hundreds of thousands or millions of dollars to develop a single prototype. At those costs, don't expect to see real world deployment any time soon. Furthermore, integrating the animatronic, structural, gyroscopic, and intelligence aspects would prove to be a major challenge, not just because of the technical hurdles of fitting all that technology into a shape/size that is mobile, and not just because of cost, but because of the intellectual property challenges that each contribution to the project would raise. Furthermore, there are lots of safety/trust issues- people would probably fear real robots/androids without some safety or assurance about them. And then there are all sorts of ethical, moral, and legal issues regarding personhood.

I think the technological hurdles will be overcome way before the practical hurdles are. You may be right, it might be a singularity style event that gives rise to the AI, and the accelerating pace of technological change, or even an AI itself, may play a role in crafting much more realistic robots. But to me, even if those things were to happen, if AI popped up tomorrow, I don't see robots inhabiting our daily lives in anything like 10 years.

But you're right, time will tell.
[User Picture]
From:ripebastard
Date:October 2nd, 2006 10:15 pm (UTC)
(Link)
I think that is quite a significant holdup... but ten years is a long time in terms of science/research/etc.

Would it really that much of a hold up? Do Andriod really need true intelligence for what we need them to do.

I'm sure given enough image recognition, mobility, and object grasping ability we could have robots work at a fast food resturant and preform simple tasks.

That or order one to enter a room and shoot anyone hold an identified weapon...

Of course that might have a sticking point...
[User Picture]
From:baikonur
Date:October 3rd, 2006 05:22 am (UTC)
(Link)
Of course robots could be used to automate simple tasks, like taking orders at a fast food place- but why? You can pay some 3 or 4 teenagers $5 an hour to do that job, or buy a million dollar robot to do the same thing. That is an investment that never yields a return. And does yield substantial legal and political trouble. Imagine the union/labor response to replacing wage slave jobs with robots.

A lot of the Japanese robots get hyped as being designed to automate certain tasks like selling movie tickets or acting as a concierge. Costs inefficiencies aside, a robot/human personality is an inefficient interface for that sort of task. Even assuming the robot performs as well as a person for that task, it is still a lot easier, faster, and more convenient to swipe your card at a Fandango enabled ATM machine or some such device rather than talk to someone.

I've said it before and I will say it again- until costs come way down and performance goes way up, these kinds ofrobots will remain animatronic novelties. That doesn't mean they won't be deployed, but they will only be deployed in places where the novelty factor makes up for the cost inefficiencies, like in resorts, casinos, theme parks, and the like.

And also, lets face it- these robots creep people out. That will be a factor that will have to be considered as well.

I don't mean to be a doomsayer, I want to see robotics, especially AI inhabited robotics flourish. but I think it is overly optimistic to assume that simply because we possess some proficiency at some of the technical challenges, that the rest will follow quickly and that society will be quick to adapt. Maybe it will- but I have my doubts.
[User Picture]
From:ripebastard
Date:October 4th, 2006 12:09 am (UTC)
(Link)
True, but if the robot was even $50,000 it would be cheaper than have an actual minimum wage employee. (remember, they aren't paying just for the employee but for other costs invovled in hiring them and plenty of other HR junk)

But that aside... What if robots were mass producable and it only cost them $1,000 to make. Then they would be used hands down all the time.

> Go to Top
LiveJournal.com