Google’s Deep Dream shows us how simple pattern play in visuals, with some iterative dynamics, can create experiences which previously felt exclusively human, spiritual and deeply conscious. This reinforces the idea of our mind as a complex network of networks and ultimately recreatable by technology.

I think I am late to the party here, but better late than never! I read the recently released white paper by Google: PathNet: Evolution Channels Gradient Descent in Super Neural Networks and became fascinated on the conceptual simplicity of the underpinning algorithms which build narrow artificial intelligence.

These methods are teaching Google and others to recognise and interpret images to aid searching, indexing, autonomous driving and many other activities. A quick and easy to understand explanation of the fundamental workings of the algorithm’s pattern composition network can be seen in the excellent Computerphile video here.

The amazing thing I have found about this form of visual based AI is how humbling it has been to the interpretation of my experiences on hallucinogens. At the time I am totally enthralled, immersed and intrigued by the visual effect of such chemicals. I spent years looking for truths and explanations in what I saw, I searched for representation from artists and contemporaries. I knew it was my mind creating the experiences with the assistance of the booster shot of chemicals but combining the visual experience with the emotional creates an otherworldy sense of being. I can easily see susceptible or maleable minds being swayed to religion and other out-there theories of extra dimensions and the such.

Well let me introduce Google’s Deep Dream: The best psychedelic visuals ever created. Not by people who have experienced them, who have been immersed within them, who try to decipher them whilst in the midst of a trip - and then attempt to remember the experience afterwards through art, but by code. An algorithm which has never had conscious experiences, either sober or on psychedelics.

Citec Deep Dream Loop by Alexander Neumann, where “… an image of the CITEC building in Bielefeld processed by Google’s neural network Deep Dream. The output is zoomed and fed back to the network as a new input a hundred times.” And here is the ultimate:

A Deep Dream 360 degree camera / VR experience! Get the mp4 for your Oculus/Vive/OtherVR.

Want to create your own? Here you go: Deep Dream Generator

Here is a paper showing you the algorithms filling in total gaps of information effectively

Here you can try drawing your own images of cats to let Tensorflow fill it in.