Since Google decided to open-source their Deep Dream visualization tool, which was designed to help them understand how neural networks work in recognizing images through different ‘layers’ of appreciation, artists and programmers all over the world have been having a blast using it to create pictures worthy of a Hunter S. Thompson’s nightmare –like the one above, featuring renowned graphic-novel writer Alan Moore.
I’ve been thinking a lot about DeepDream lately, and what it could possibly mean. Is it a quirk of Google’s algorithms, or is it hinting at some fundamental aspect in the process of perception itself?
And, could it be related somehow to the UFO phenomenon aswell? I think it just might! I think in fact it’s the perfect example of what my friend and colleague Greg Bishop has been proposing for sometime, of how when we observe something completely outside of our normal frame of experience –say, a disc-shaped object in the sky, or a humanoid form on the ground– we unconsciously struggle to put inside one of our ‘mental boxes’.
When we finally make it fit, the end result might be an aberration from the original signal, shaped by the expectations and desires of the observer.
In any case, DeepDream should at least give us some comfort in thinking the Robot Apocalypse is gonna be groovier than we thought –and to stay away from the killer sentries, just hide behind a flock of chickens…