It is finally sunny today after many hours of fitful, glowering weather that could not decide if it wanted to rain on us or not. It is the sort of weather that makes for maximum suspense if you pay attention to that sort of thing, because it literally feels like anything could happen. It could rain. It could become all picture postcard sunny. Flying ants could descend upon us in hordes and force us to learn calculus. Really, anything could happen.
But neither rain nor sun nor mathematically inclined formicae shall stay us from our appointed rounds! It is time for the science, bitches, and I hope you brought your wigs, because they are about to be flipped.
As Simple As Black And White
We will lead off today with a truly freaky cool (and totally creepy) bit of technology from those wonderful people in the covert surveillance business.
You know, spy stuff.
An outfit called Qwonn Security Products has come up with a material they call Black Ops Plastics that has the remarkable property of being completely opaque – black, even – to the human eye, but invisible to a black and white camera.
That is right, a black and white camera can see right through it. This means that you could hide a black and white camera inside any object made of this amazing material, and it could take footage without even so much as the tiniest lens showing.
Or you could have a whole wall made of this stuff, and be watching people via black and white cameras all along the wall without them knowing.
All in all, it is a big boon for covert surveillance, whether you are a professional spy looking to find out what a terrorist is doing, or merely an amateur pervert looking to find out what really goes on in the ladies’ room.
It’s A Quantum Finish
Next up, we have a story that present an interesting problem : what happens when instant replay tells us there is no winner.
And this is not just any instant replay. Let me explain.
What happened is that last weekend, two sprinters aiming for a spot on the American Olympic track team – Allyson Felix and Janubah Tarmoh – both finished their race at exactly 11.068 and no amount of analysis of the photographic results could determined which of the two placed third and which placed fourth.
Small problem : there are only three slots on the team.
Oh, and here is the kicker : the camera used to record these events and determine result records at a whopping three thousand frames per second, or 3000FPS for you camera nerds.
Imagine that. Three thousand frames a second and none of them show a clear winner. Common sense would say that was impossible, but it’s not. It is just incredibly unlikely. Enough heats in enough events, it was bound to happen eventually.
By the way, The International Olympic Committee decided that the result would be determined by either a tie-breaker race, or a coin toss, athletes’ choice.
I feel bad for the two ladies either way.
Intelligence and Cats
Meanwhile, those great big boffo brains at Google have been doing what any group of billionaire nerds with massive amounts of resources would be trying to do : make the world’s first artificial intelligence out of a neural network of 16,000 computer processors.
A neural network is one in which the individual machines and the connections between them are modeled as closely as possible on the human brain, with each machine a neuron.
The article is annoying vague about what, exactly, they did with said network, but basically, they fed it ten billion images taken from the frames of YouTube videos, and the network trained itself (?) to recognize various objects within those images.
I do not quite fully comprehend, but I am glad that their results seem to confirm the theory that the brain uses specific neurons to recognize specific objects.
But we all know what they were really doing : testing the theory that if you create enough connections between enough neurons and then give the resulting network literally any kind of stimulus, consciousness will spontaneously arise.
Guess 16,000 processor-neurons was not enough. Next time try a million!
Almost a Turing
In a normal week, that would be a pretty amazing result in the world of artificial intelligence. But Google’s intriguing experiment pales in comparison to the news from last Saturday.
Last Saturday was June 23, 2012, and would have been the one hundredth birthday of computer scientist and all around amazing meganerd Alan Turing 100th birthday.
To celebrate, they held the most massive Turing Test event ever, and as is befitting Turing’s centennial, a chatbot named Eugene Goostman won the competition with a nearly 29 percent result.
To refresh your memory, the Turing Test states that if a computer program can fool thirty percent of the judges into thinking it is a human being, it is to be considered artificially intelligent.
So for Eugene Goostman, a chatbot that simulates a 13 year old boy, to come so close to Turing’s target is truly enormous news.
The fact that it happened at an event celebrating Turing just makes it all the sweeter.
The article is sketchy on to how it pulled this off, or even who the hell programmed it, although if you want to talk to Eugene, you can do it here. That would seem to indicate that it was the Princeton AI team who programmed Eugene.
Way to drop the ball there, Pop Sci. Good thing you don’t cover sports or the headline would be “Some Guy Beats Home Run Record”.
But one clue as to their technique might be their choice of a thirteen year old boy as their persona. That way, the chatbot can appear intelligent, but any lack of verbal sophistication will be explained away by observers as, well, he is only thirteen years old.
In fact, in the future, using a sub-adult persona might well be deemed to be cheating. I am sure that simulated the text input of a toddler would be simplicity itself.
Just a random number generator plugged into the ASCII table, really.
See you next week, folks!