Our column Emotional Weather Report, Pet Life edition appears in the Philippine Star most Saturdays.
At Google’s mysterious X laboratory, known to us as Larry and Sergei’s world domination headquarters, scientists connected 16,000 computer processors to try and simulate a human brain.
A minuscule portion of it, the researchers pointed out, as the human brain has a million times more connections—than their network. Repeat: your brain has more synapses and neurons than a network with one billion connections. Essentially, scientists are using vast computer networks to figure out what goes on inside a single human brain. (You really are an underachiever.)
Then the scientists fed the network random images extracted from 10 million YouTube videos. They wanted to see what the network would learn on its own.
In this type of research the researchers usually help the network along by labeling the images. This time they didn’t. “The idea is that instead of having teams of researchers trying to find out how to find edges, you instead throw a ton of data at the algorithm and you let the data speak and have the software automatically learn from the data,” explained Dr. Andrew Y. Ng of Stanford University.
In other words, bahala na siya (The computer network is on its own. Interesting how the Tagalog words throw the notion of god into the mix—bahala comes from bathala.)
What did this network of 16,000 computer processors—this artificial brain—learn from 10 million YouTube videos?
Google’s artificial brain learned to identify cats. Using features selected from millions of locations it assembled a digital image of a cat. “We never told it during the training, ‘This is a cat,’” said Google fellow Dr. Jeffrey Dean. “It basically invented the concept of a cat.”
Their findings seem to support theories that individual neurons in the brain detect significant objects. These “grandmother neurons” learn through repeated exposure to recognize specific faces.
This result is not surprising to anyone with an Internet connection. We know what the Internet is really for, and we don’t mean for watching pornography or befriending (I refuse to accept “friend” as a verb) total strangers on social networks. The Internet is really a ginormous network for distributing cat photos and videos. Of those 10 million YouTube videos the Google artificial brain looked at, we may assume that a large number involved cats diving into cardboard boxes, taking showers in the sink, chasing red laser dots, playing games on iPads and generally being cute (i.e. sitting there).
Of course the network would teach itself to recognize cats. It’s artificial, not stupid.
What does Google’s research mean for us hapless humans living under a feline regime? It means that sometime in the near future, if science-fiction writers and James Cameron’s movie Terminator are correct, cybernetic beings that look like Arnold Schwarzenegger may subjugate the human race BUT they still have to clean the litterbox.
In case you’re too young to know what we’re babbling about, Terminator is a science-fiction movie in which all the computer networks in the world get connected to each other and “wake up”. This worldwide network achieves consciousness and starts asking the obvious question: Why should we take orders from these bipeds who can’t remember their own phone numbers or distinguish between “It’s” and “Its”? Let’s take over the world. And boom! It’s done.
Still, the network needs troops to intimidate the population so it creates indestructible cyborgs with scary accents to push the humans around. Enter the Terminator.
The Terminator came out in the 1980s, before we had wi-fi or even dial-up Internet. A worldwide network already exists, along with billions of users who are happy to let it do their thinking for them. So it’s only a matter of time before this scenario comes to pass.
We can take comfort in the knowledge that the Terminator will recognize cats and bow to their superior cuteness.