Slashdot Mirror
Sony's Robot Attends Pre-School
Darren writes "Sony's Qrio humanoid robot has been attending a Californian pre school to play with children under the age of 2 since March to test if robots can live harmoniously with humans. I wonder if the testing includes monitoring the 'nightmare status' of the pre-schoolers?"
Humans are driven by various needs (e.g. shelter/sex/food/beer) - what needs do the robots have?
The driving interest in toddlers (and that's what the article is about) certainly isn't sex or beer, and it also isn't shelter or fod - which is still provided by the parents.
The driving interest in very young kids is pure interest. Our brains are just wired that way. Curiosity is a built-in feature.
Assorted stuff I do sometimes: Lemuria.org
I'd bet that the first human-equivalent machine intelligence takes 18 years to develop after the first human-brainpower-equivalent CPU is created. It will take that long for the machine to "learn" the world if it only has a CPU equivalent to one human brain (1 HBE).
Of course, if Moore's Law is still kicking, then 2 years into the learning phase, they can swap the 1-HBE processor for a 2-HBE processor. This will shorten the remaining learning period, but I doubt it will cut it in half. Learning to physically and mentally interact with the world will still take time. What might accelerate the learning time is if multiple copies of the intelligence can share experiences and learn directly from each other's mistakes/successes.
The point is that the first intelligent robots will need to go to preschool to learn how to interact with the world.
Two wrongs don't make a right, but three lefts do.
I wonder if the testing includes monitoring the 'nightmare status' of the pre-schoolers?"
I wonder if the submitter has any clue as to what he's talking about.
It's pretty difficult to give toddlers nightmares. They're not easily scared. They do cry over the slightest problem, mostly because crying is the only well-developed form of verbal communication available to them at that age. They are also excellent at forgetting whatever the problem was and getting on with their lifes. Watch a kid hurt itself. Then go away and watch the same kid 10 minutes later.
It'd take a serious event to cause nightmares in those kids, and that machine has neither the looks nor the sheer physical power that would be required.
Assorted stuff I do sometimes: Lemuria.org
I always wondered what motivation robots have for "learning".
Robots have no motivation other than that given them by their creators.
Robots are not sentient. We do not even know what sentience is. The only way for us humans to create sentience is to procreate.
what needs do the robots have?
Errm.. like all machines, they need a power source. That is all.
Talking about robots as if they are alive and have motivation other than their code implements belies your otaku sensibilities. Clearly, you have not yet procreated, or you would not be so obsessed with making a machine which 'pretends to make it look as if you have done so, technologically'.
; -- the corruption of government starts with its secrets. a truly free people keep no secrets. --
Robots have no "motivation" to do anything. they have a reward function that they try to maximize, but certainly it's not anything like that capricious human thing we call "motivation" (which is actually a very good thing).
Again, it should be mentioned that while it may make us feel very cool and cutting edge to apply human terms like learning, thinking, or motivation to machines; they really are ultimately meaningless in a non-human context and are only useful as analogues and in impressing your grand-mother with how her tivo "learns" her tastes
as Edsger Dijkstra famously said:
~AS
I think they'll find that it's not a matter of familiarity. It's a survival reflex and it's pretty deep. Your brain flags "almost human" things as grotesque and something to be avoided. It's why many people are afraid of clowns and wax figures. They look almost human, but still look wrong.
People would be far more comfortable with Bender-like robots than with "I, Robot" style robots because they don't try to be human, just humanoid. If it looks sufficiently non-human to avoid triggering that reflex, they'll be alright. Other than that it'd have to be completely perfect, like Data.
Never confuse volume with power.
or maybe 9 years if we take into account that we need to rest (even though part of resting this time is important with regard to the learning activity)
You may be right. The question is: is sleep/relaxation, etc. a critical part of intellectual development? For humans it definitely is -- sleep deprivation really messes up the brain. But even for non-biological intelligences I'd bet that some "downtime" is an important part of assimilating all the data of the day. Interacting with the world is a full-time job for the CPU that forces the deferral of many analysis and restructuring tasks that can only occur when the brain is offline.
Perhaps androids would dream because dreaming is a critical maintenance/analysis cron job.
Two wrongs don't make a right, but three lefts do.
Yes, but curiosity killed the cat...
Seriously, it's easy to get led astray using evolutionary paradigms to explain traits. We often think of something as a clearcut, atomic quality that benefits or harms the individual.
Curiosity is a good example. Clearly in an organism whose survival depends on complex and learned behaviors, a certain amount of curiosity is needed. But most people grow out of it and become dull,predictable, dependable adults. But some don't -- there's a continuum. And the variance of that trait in adults is useful to the tribe, if often harmful to the individuals on the right end of the bell curve.
Og: This flint is mammoth dung! It keeps shattering when I try to work the edge.
Gog: It's good enough. Just chip another piece of and sooner or later you'll get a good one.
Og: Crap. I'm going to find some decent flint. See you in a few weeks.
Now it may be frequently that Og comes up empty, or is killed, or gets lost and never re. Og is the type who runs across a cave and finds it impossible not to explore it. Now he risks getting eaten by a cave bear, but when he doesn't get eaten, he may have found the tribe a place to hide in times of trouble. The tribe benefits by having a few geeky cavemen and -women who can't keep their nose out of trouble, and the risk is concentrated on a few individuals, whose types will be reproduced again by the future variation in the trait.
I think this is one fundamental difference between robots and humans. Being a human is like playing a game in which you don't really know the cards you've been dealt or are playing, but have to infer what's going on by how the play goes. Being a human is a journey of self-discovery. To design a human robot, you'd have to make it ignorant of it's own characteristics and make it have to deal with the consequences. Until that happens, a robot is just going to be an object.
Post may contain irony: discontinue use if experiencing mood swings, nausea or elevated blood pressure.