As seen at NewMobileComputing: Meet Cog, an aluminum robot born in the MIT Artificial Intelligence Laboratory and a compelling experiment in robot cognition. It’s tethered to dozens of rack-mounted PCs running QNX, the popular embedded systems OS.
Robotics: Cognitive Machines
About The Author
Eugenia Loli
Ex-programmer, ex-editor in chief at OSNews.com, now a visual artist/filmmaker.
Follow me on Twitter @EugeniaLoli
Robots are great for assembling cars. Thinking that you can program a robot even to the intelligence of a three-year-old is a bit much. I am not sure they will ever intellectually match up nor do I want them to intellectually match up with my three year old.
That’s simulated intelligence
While I’m certain there’s a long way to go, “ever” is a long time. For me it’s just a matter of “when”. I predict that we will have adult human level intelligence AI in 30 years.
The key to getting it that soon is that each experiement will build not only on previous experiments, but will also increase the rate at which we can learn about human intelligence. We’ve got to “ramp up” core technology to get there — processing speed and so on — but I don’t think that’s the bottleneck. The bottleneck is, I think, that we really don’t understand human intelligence. But by trying out various simulations over the years will reveal as much about human intelligence as about the potential of AI. We’ll know that our simulation is a good approxomation of reality, and therefore our model is a good model, when our simulations begin to resemble reality. Basically I consider human intelligence complicated enough where no model in English will be able to express it (“ever” — there ya go! :p) but a description using a language designed to express process — programming languages — probably can.
Erik
Well, it was just a suggestion. There is NTFS support being developed by a russian progammer at qnx.org.ru.
Support for ReiserFS on QNX would be nice, and there is nothing wrong in suggesting it!
Take it easy on the moderate button, me lady.
-Magg
MIT gave up their AI project a few years back. They weren’t making significant progress to justify the vast amounts of money they were investing in it. Around the same time a group in the Netherlands? had to give up their project for AI as well as the goverment refused to fund them any further.
AI or “cognition in robots” or whatever you want to call it seems to me to be a waste of resources because you just know when they perfect the robot worker, the robots will start a union and hold strikes
.
“AI or “cognition in robots” or whatever you want to call it seems to me to be a waste of resources because you just know when they perfect the robot worker, the robots will start a union and hold strikes
.”
Not necessarily. You could wire the robots is such a way that they’d get enourmes satisfaction, or better yet, get high from working and being treated like crap. Though I suppose these robots might get dangerous when there’s no work to do. “Give me work.. must get HIGH!”
IMHO AI is not going to happen until processors are
big and complex enough not to be understood by human
beings… if we can’t understand our own brains,
how can we expect to undestand or to create intelligence
using techniques that fit in a 500-page book?
IMHO, the memory is the problem.
While I’m certain there’s a long way to go, “ever” is a long time. For me it’s just a matter of “when”. I predict that we will have adult human level intelligence AI in 30 years.
No way. I actually studied AI this past semester. The really substantial advancements have been made in neural nets. A truely intelligent AI agent in my opinion would have to be designed with a neural net, since a neural net works very very similar to the human brain. I don’t believe this is the ONLY way that an intelligent agent can be designed, but regardless, there’s no way that we will have a truely “intelligent” agent within 30 years. It’s actually hard for anyone to make a good prediction about this, because as Alistair said, there has already been one fallout point where everything was stopped do to poor progress. Maybe that will end up happened again; who knows. If I had to take a guess, I would say no less than 50 years for a truely intelligent agent.
I predict that we will have adult human level intelligence AI in 30 years.
Funny thing is other people were predicting this 30 (or more) years ago – remember HAL in 2001 – released in 1968.
It could be that intelligence is an “emergent behaviour” that can’t be pogrammed, it just needs the right enrionment and it’ll just appear.
Are we creating the right envrionment though? Probably not – artifical neurons do not model real neurons exactly and they certainly don’t replicate the electromagnetic or quantum effects within the brain. But maybe they are not even, important who knows…
so why can’t a robot collect information, see how it relates to other information it has, then draw conclusions (as every child does and every adult does even better) based on the relationships?
that is all inteligence is. why are we putting such a high pillar on human inteligence and acting like it is impossable to create something mechanicel that is cappable of the same?
While were on the subject of MIT robots check out the MIT Leg Labs latest robot:
http://www.ai.mit.edu/projects/leglab/robots/robots.html
We put such high regard towards human and other organic intelligence, because it is sentient and aware of it’s surroundings. It can in effect break out of it’s genetic “code” and do the unexpected or indeed previously thought impossible.
AI would need to be able to do things it hasn’t been programmed to do i.e. program and enhance itself to be anything like humans.
sfd;jdjk;fkjl dkljfksdfsa