The good thing about little brothers is that they get awesome toys for Christmas, and that sometimes they share.

My brother got a lego robotics set as an early Christmas present, and he’s really into making robot protectors.  The first one shoots balls at anyone that comes too close, and the second is a robogator that pounces and bites if you come too close.

They don’t give their maker any special treatment though:

photos from Vogue Girl Korea, posted by cathastrophe

I think about love every now and then.  I think about robots a lot.  Sometimes i think about both of these at the same time.

Sometimes it’s because I come across articles like “technosexual” and “companion robot aiko” that fill my heart with both pity (With the current state of robots right now, the “love” returned by the companion robot is arguably less than returned by a pet fish) and understanding (It’s pretty tough to find someone to love, and if they are happy this way, why not? Plenty of people love jerks who return less love to them than their pet fish does.)

Sometimes it’s because of photoshoots like this.

Sometimes it’s because of watching WALL-E, and concluding that it might be one of the best love stories of this decade.

Other times it is musing about whether or not I would build the 100% perfect robotic man for myself, if humanoid robots become indistinguishable from humans in our lifetime.

Okay, actually, I think the idea of building humanoid robots that are indistinguishable from humans is kind of dumb, since… existing ones are easy enough to come by, and many could probably use more attention than they’re currently getting.  The amount of effort needed to make robots indistinguishable from humans will be excruciating (getting non-linear joint motions to be perfectly fluid, perfect texture of skin, non-creepy facial expressions, and we know from the Uncanny Valley theory, it’ll get worse before it gets right) and with very little return value.  Wouldn’t that effort be better spent developing AI to parse through genomics data to cure diseases and economics data to solve world hunger?  These tasks don’t require looking human.  There are other “precautionary” reasons I am against it.  If robots are pretty much indistinguishable from humans, with the same volatile emotions and ego, except possibly with crazy super powers (like guns built into their arms) and crazy super cleverness, well, you can imagine what might happen from the 20+ robots-take-over-the-world kind-of-plot movie/books/comics.*

That last point is probably more of an indicator that we shouldn’t try to build machines that inwardly resemble humans.  There’s a saying that no writer can write a character smarter than himself or herself.  But I think we can make machines smarter than ourselves, if smarter means “makes better decisions”.  People’s decision-making is usually hindered by time, by short-sightedness, and by emotions.  If a computer can quickly process many levels of a possibility tree for all the possible available actions, that’ll help them make better split-second decisions than people can.  Understanding emotions might be necessary for making some kinds of decisions (like how best to introduce a sensitive subject in a conversation), so robots that interact with people will most certainly need to understand emotions.  And maybe complete stoicism would make for creepy human-robot interactions.  But perhaps a “wall” could be built to prevent emotions from overly influencing humanoid robot decisions– like a Markov model, many points of view would be considered in decision making, and the emotional response of a robot would have only a small weight.  At this point, please don’t jump on me for devaluing what it means to be human.  In fact, I kind of like being human; when my heart gets stomped on sometimes I revel in the fact that I have the capacity to feel aching– I savor the tangibleness of my pain in the way I savor the acute sharpness of a massage; my biggest fear in fact is to become so jaded as to feel nothing.  But the most emotionally mature of us treat emotions as a code of our and others’ deepest needs, and then act on the messages in these codes instead of being overwhelmed by them.  I accept and embrace my moments of emotional overload, because, they are real feelings to me, and it’s part of being human, and what else can i do about it?, but as long as we’re starting from scratch, machines in charge of making important decisions should not freak out, be jealous, or throw tantrums.

But, despite my banner-waving protests and warnings, thanks to tons of weirdos researchers who think this will be very marketable or want one for themselves, companion robots are for most purposes nearly indistinguishable from humans inside and out will exist someday.  And, as long as they are around, will I refrain in silent protest on principle, or march to the store with that 5-page list of ideal traits i wrote up in the 9th grade?

The 100% perfect robotic man.  100%.  It’s pretty tempting.

Ack, actually, my model is in no way inspired by internet dating.  In this day and age, I am not ashamed to say that I have gone on some “dates” with people i have met online, but I can count them on one hand.  When I say 80% match, it is roughly approximated from how comfortable, happy, and magical a person makes me feel.  And when i say 100% match, I don’t mean someone flawless– I like my men with foibles that are simultaneously irritating, hilarious, and endearing, after all.  When online dating sites say I match 80% with someone, they mean by interests and easily categorized beliefs.  Okcupid pretty much says I match 90% with everyone.  This is useless to me, and I just ignore it.  Instead, I read what someone has to say, and if I have something to say back, I do.  It’s like leaving a comment on a blog.  Or it’s like saying “Hey, have you been to the new Yoshimoto Nara exhibit?” to the guy reading “SUPERFLAT” and sipping a matcha latte in a wireless cafe, except he’s in a wireless cafe on the other side of town.  I conceived the match-percentage model a year or two before I tried internet dating; it was born of thinking about statistics, games involving chance, and algorithms that store a temporary best approximate solution but keep refining it in the background.  But really, “conceived” is way too fancy of a word– I hardly thought about it at all, and you could come to the same conclusions just by observing your relationships and your friends’.

Internet dating and tricksy genies

However, internet dating did reinforce something to me about life and dating.  I think of it as the tricksy genie phenomenon.  You know, how if you ask a mean-spirited clever genie for anything, he’ll give you what you asked for without giving you what you actually want, by taking advantage of fuzzy details in your phrasing.  That’s what most online dates have been for me: dinners with boys who sound pretty cool in theory, but with whom I mysteriously share zero chemistry.  Maybe that’s how it would go with Mr. 100% Perfect Robot.

Some people think robots will save marriage in Japan

Some people look to robots as a savior of love, even if not a source of love. This interesting paper talks about why Japan is leading the world in resources devoted to robot development, and it has to do partly with Japan’s marriage/family crisis.  Basically, for various reasons many women are not getting married, and many married women are not having children.  One oft-given reason is that these women would like to have more time to focus on their own careers and goals.  A government conference was held, and a few solutions were proposed including: robots that help with housework.  It’s a bit ironic really, to think that some men work late hours at night on robots that will help their wives do the housework, hours that they could be at home helping out with the kids and the dishes.  But I guess, stick to what you know.

Robot lovers as the world’s jealousy/envy remover

Elaborate plans to find love, elaborate plans to keep love, we put so much effort into love!  Robots will make people lazy in many many ways, if they do our building and cleaning and even researching, but if robots could guarantee us love by being perfect mates for us, that might result in the ultimate laziness.  Imagine it: a robot that was 100% perfect for you and loved you exactly as you are.  Ideally, people work hard to achieve their goals so they can be happy and proud of themselves.  But if people knew that the perfect lover would still take them and love them forever lazy, grouchy, and rude, maybe some people wouldn’t strive to better themselves at all.  On the other hand, guaranteed robot love would probably prevent jealousy, envy, and fighting.  It would be a peaceful world full of terrible, lazy people.

In the End

The future robot-occupied world which i suppose i am contributing to sounds amazing and boring and scary all at once.  I guess it is good and bad that it is a long way coming.  We are working towards a disease free world where everyone gets fed and loved, but at the moment i might even prefer some of the miseries.  I suppose i am not even particularly attached on the ideal boy in my mind because I trust that this crazy universe is far more creative than me, and I would love to fall in love with someone accidentally for reasons I would never imagine.  And so I am pretty sure that I would never custom order a robo-lover, but if I one day crossed trajectories with a charming 90% perfect boy to find that he’s a robot, well– then I’d ask him to help me build a rocket and we’d explore the universe forever.

* Maybe in some future post i will discuss my elaborate theories for how to prevent world take over by robots.

I wanted to link to this breakthrough robot news story because quite a few of my coworkers are on this project– and the top photo of the article features my friend! (the one on the left)

Basically, using a sensor that detects small electric pulses on your scalp, and infrared sensors that detect changes in your brain’s blood flow, and analyzing patterns from many test cases, the Brain/Machine interface can differentiate between certain simple thoughts, such as, thinking about moving your left or right hand. and the Brain/Machine interface is linked to an ASIMO robot, which will move his corresponding hand. You control ASIMO for simple tasks with your BRAAAAAAINNN (cue creepy music).

It’s scary and exciting huh! It sounds so science-fiction-y but it’s true! (though in it’s infancy)  I’m thinking very Paprika thoughts, and wonder if one day we will be able to have intimate mind-to-mind conversations with other people.  Or record our dreams. As for mind-controlled driving, though, i’d much much rather that a smart car just drove itself.

You can read a more informative account of the Brain/Machine interface and watch a short movie here.

i got to listen in on the presentations of honda student interns that are just now graduating.  i didn’t understand too much because, it was in japanese, and so i had to make do with my limited techincal vocabulary and slide pictures.  still, i’m probably not allowed to say very much about these things so this keeps me honest.

- ASIMO turning a jump rope with a human (matching the force of the human so as to swing the rope evenly).  this is significant because the wrist movements involved form a non-linear path.

- ASIMO picking up a beat (any kind- music or clapping) and stomping and singing in the correct time.

some voice recognition stuff:

- echo removal for voice recognition in an echo-y room

- channel splitting of many voices into individual voices (voice recognition in crowded areas)

a friend theorized to me a while back that japan will probably be first to perfect voice recognition technology because of the small number of fixed syllables, and the existance of fixed “appropriate” responses in many office situations.  though, i guess because of the small number of fixed syllables there are also lots of homonyms that have to be figured out from context.  still, i think he’s right.

these projects are all still in the realm of disjunct ideas that don’t constitute any kind of “intelligence”, even when all put together.  but they’re cool and potentially useful nonetheless.

I’ve been reading “The Emotion Machine” by Marvin Minsky and I’m enjoying it, a lot.  I am sort of ashamed that I missed out on the chance to take his class.  There are so many texts that miss the point of AI, I think; I read many such misguided texts and I wish I had discovered Marvin Minsky in highschool.

Honda’s vision is to make a robot that can move and interact with the world and give it enough intelligence that it can build up knowledge about the world from those interactions, like children do.  While there are other research groups with similar goals, I’d argue that Asimo is the cutest humanoid robot.  The first time I had to perform some tests on him was the first time I saw him up close.  I seriously wanted to hug him.  He’s small, with big feet, and large “eyes”.  When he walks around, he turns his head first, making him look a little confused.  He has a frame that covers his wires but he’s not human-looking enough that he falls into the uncanny valley of creepiness.  And I’d also argue that cuteness will be a big factor in the success of this kind of robot.

I’m being totally academic and serious here.  Ok, so it’s no secret that I am an overzealous fan of cute.  And if I had it my way, everything in the world would be cute, from toilets to trains to garbage cans, whether it needed to be cute or not.  Wait.  Isn’t cute for cute’s sake?  Does anything need to be cute?  Most times cute is just an embelishment or a marketing ploy.  But yes, actually, there are times when cute serves a purpose.  Children need to be cute.  Think about how annoying they are.  Imagine if an adult had the behavior of a child.  Picking his nose, screaming in public, pulling on your sleeve every 5 minutes to ask you to tie his shoe.  But we love children!  Children get away with this mostly because they’re cute; the urge to fawn over and caress and protect anything with baby-like proportions is surely evolutionarily beneficial.  We put up with the ridiculous demands of children because of their doe eyes.  We refrain from strangling them because their dimpled roundy faces.  They act inappropriately because it’s their way of testing hypotheses.  And they pester us with questions because they need others to help them learn.  Just like a learning humanoid robot will need frequent affirmation for the conclusions it’s drawn, just like a learning humanoid robot will break your favorite coffee mug, ask you rude questions, and need repairing when it falls.  And imagine getting pestered by Creepydoll McSandbags here.  I don’t think I’d answer his questions patiently in a manner conducive to his learning. I’d answer quickly and leave!

Asimo on the other hand, looks like a child explorer from an alien planet, complete with robotic back-pack, here to gather information for his 3rd grade class project about the earth.  How could you not be happy for him as he figures out how to navigate the building, or feel compelled to help him out when he’s a bit stuck?  A cute humanoid robot is going to be the most successful at acquiring knowledge like a child, because people will enthusiastically and naturally interact with it closer to the way they’d interact with children.