Last night I nerded out at AWE-MAGEDDON*, a live event series put on by my favorite WNYC podcast, Radiolab. The show featured Brooklyn-based duo Buke and Gass, creating the sound of five musicians on their homemade baritone ukelele and guitar-bass hybrid -- as well as robotics engineer Hod Lipson, whose work in evolutionary robotics involves creating machines that can adapt and demonstrate elements of human creativity (and take over the world...?). Nerd central! Professor Lipson introduced us to one of his four-legged creations that has been charged with the instruction to move forward. But this robot does not know that it is four-legged. It does not know anything about its nature, in fact; it will create a computer simulation of its sensory input as it exists in the world, and then figure out how to walk from that simulation. Lipson tells us that the robot starts by making a random motion, assessing the sensory input that it receives from that motion, and then it begins hypothesizing as to what its shape and form might be.
Here's where it gets interesting (as if it weren't already). To test the hypotheses, Lipson says, the robot starts looking for "the most disagreement". It tries movements specifically to rule out one possibility or another. It tips and tilts and tries to throw itself off its axis, because that will give it unequivocal information about what it is and what will or will not work. It looks, in other words, for failure -- because that is the fastest way to learn.
There is a lot of lipservice paid to failure -- don't we all know by now that you learn more from your failures than your successes, and everyone falls down a few times on the way to the top? But - and maybe it's just perfectionists like me - but I don't think too many of us go LOOKING for failure.
I certainly hope it's true that machines will never be able to embody the human creative spirit, our ephemeral impulses and nuances and subtleties. But maybe there is something to learn from this non-human creation, with no emotions and no attachments and no desires, and no need to have great PR.
Because it is entirely possible that the fastest and most useful information about my nature; about what and who I am; about how to move forward... could all be found by throwing myself off the cliff and hoping for an EPIC FAIL.
*You can watch the event webcast here!