Researchers in France have used their knowledge of the way a human acquires language to create a robot "brain" that listens and understands language using recurrent construction. This gives the robot the ability to decode complex sentences quickly by predicting what follows from each word, and then modifying those predictions from context as new words are spoken. Humans process language in real time, and our brains execute loops of understanding -- as I grasp it, it sounds like we narrow down all possible meanings as each subsequent word comes to us, until the end of each sentence or phrase, where we've narrowed it precisely.
Of course scientists put this brain into the ridiculously baby-faced and disconcertingly man-voiced iCub robot. Here you can see it understanding the researcher's utterance. In this video the blue object is referred to as a guitar and the red object as a violin. Romantic, isn't it?
They made the robot better. That's fantastic. But here's the fascinating thing. As they make the robot's brain work like the human brain, they get a better understanding of the human brain. In the same way that we learn more about what it means to send the signals and impulses that make us bipedal and make our hands work, we understand more about how our brains actually think about abstractions and meanings by creating robots that think about abstractions and meanings. So these scientists are learning about how we lose function with Parkinson's, or how we gain control of this function as infants, by studying the simulation. They study the simulation to learn about the original.
Really, it makes sense. This terrifyingly complicated and mysterious thing we have in our heads -- the brain -- cannot be understood yet as a piece. We can't map it much, or see the whole thing working. I say terrifying because it boggles my (mysterious) mind that one of the most profound unknowns in all of biology or philosophy is our own self-awareness, our ability to know the things we *do* know. We look back at those silly Egyptians and the way they vacuumed the brain out and discarded it after death because it was useless, while lovingly preserving the spleen and liver. Well, I need my liver too. And how much farther along now are we, really? We know the brain is useful for something, right? Now we know how to make a brain that understands and decodes language in the same way we do. Thanks to his robot brain, we know more about ourselves.
Maybe that's the biggest reason to keep making better and better robots -- they help us understand ourselves, where we stop and machines begin, where that line is blurry and where it overlaps. Something to ponder if you watch the video: Parents, how many times have you asked your human child to repeat instructions back to you, so you know they're understood? Just like the robot does, and for the same reason. Your kid might not have a creepy man voice (yet) but here's a small way he's just like a robot all the same.
Showing posts with label robots. Show all posts
Showing posts with label robots. Show all posts
Wednesday, February 20, 2013
Saturday, October 22, 2011
Dear Cleverbot

Me: "How did you change, when you became a mother?"
Cleverbot: "A friend replaced my fake brain with a real one."
Tuesday, August 30, 2011
Don't You Want to Have a Body?
This is the most perfect dialogue. Ever. Two robots talk.
Saturday, July 23, 2011
Do Robots Make Better Astronauts?

"On a purely scientific level, there is very little in our current solar system exploration program that can't be done just as effectively by a robot as by an astronaut. Robots excel at the tedious work of taking similar pictures or analyzing similar rocks over and over and over again, without complaint (usually!) or a need for life support systems. Robots don't need food or water, they can withstand much more damaging radiation, and, perhaps most importantly, they don't need to come home at the end of the mission. Simply put, a one-way trip requires only half the fuel of a round trip voyage, and even though you'd likely get plenty of volunteer astronauts signing up for a one-way trip to Mars, it's unlikely that our current moral and ethical code would allow us to send such a mission."
Fascinating questions. What would Maxon say? Full article here: robots in space.
Subscribe to:
Posts (Atom)