As we all know, Google is the leader in search and they keep getting better. Their ‘spiders’ regularly scan every web page, analyze every word, every link, each plus one, social share, email interaction, Google wallet transaction, et cetera. Google maps continually updates its satellite data with images of every continent, country, state, building, road–even their ‘street’ views can now display underwater. Essentially, Google can get you any piece of information on the web using almost any word, image or sound.
Now imagine, for a moment, that you are not the person submitting a web search. Instead, a robot (let’s call him Hal), wirelessly connects to the Internet, perpetually surveys its surroundings at lightning fast speed while indexing information into its own storage unit and analyzing any additional data entered into its system. You hold up your coffee cup and say, “Hal, what am I holding?” Hal recognizes his name, captures the image of the coffee mug with his camera lens eyes, scans every comparable image on the Internet, finds the most relevant word for those images and in less than a quarter second echoes back, “You are holding a cup”. Now imagine Hal’s driving your car (or is the car).
Sound impossible? Over the summer, Google scientists went even further, putting 16,000 processors together and creating the ‘Google Brain‘–a machine learning neural network that taught itself to identify cats. Meaning, it didn’t just search the web for cat images and related words to help identify a cat, it “basically invented the concept of a cat,” says Google fellow Jeff Dean, who led the study.
A recent Forbes article explains how Carnegie Mellon researchers have outlined, “an artificial intelligence system that can watch and predict what a person will ‘likely’ do in the future…” DARPA, the Defense Advanced Research Projects Agency which develops new technologies for use by the military, funded this project as well as the DRC or DARPA Robotics Challenge–a public crowd-sourcing initiative for robotics research. Regina Dugan, former DARPA director, who led next-generation manufacturing and crowdsourcing hacker outreach efforts like the DRC, has been working for Google since March.
100 years from now or maybe even a few decades, it may be likely that one looks back at Google as the proverbial Cyberdyne systems, the company that created Skynet–the artificial intelligence featured in The Terminator franchise. But really, one can argue that A.I. is already here.
The familiar Google search results page you see on a daily basis is a dumbed-down version of information that a computer could process several times a second. Once these search processes are automated for words, images or sound–and improved upon with machine learning–it’s not hard to imagine this functionality embodied in a robot. When that future comes, I just hope Google has truly lived up to its motto – ‘don’t be evil’.