Teaching image-recognition algorithms to produce nightmarish hellscapes


[Read the post]

Huffing Boing Boing

That image on the main site looks like a Jim Woodring…


Is this program the source of Rob’s nightmare image?

If so, thanks for posting the paper.


Apparently it’s not one of the images they’ve posted, so…?


Machine eye movies of Jim Woodring criticism, 3D printing, masterful engraving, not a bad bunch of expectations going on here. …might be worth mentioning that click to embiggen works on the given sauce. Not seeing Giger yet. Thanks!


It seems like an ai breakthrough to me. An algorithm for dreams.


All the objects—organic and inorganic alike—were totally beyond description or even comprehension. Gilman sometimes compared the inorganic masses to prisms, labyrinths, clusters of cubes and planes, and Cyclopean buildings; and the organic things struck him variously as groups of bubbles, octopi, centipedes, living Hindoo idols, and intricate Arabesques roused into a kind of ophidian animation. Everything he saw was unspeakably menacing and horrible; and whenever one of the organic entities appeared by its motions to be noticing him, he felt a stark, hideous fright which generally jolted him awake.

– H.P Lovecraft, “The Dreams in the Witch House”


Looks like a TOOL album cover.


When I put that image from the little AI that only sees eyes on Facebook, Facebook’s AI saw faces in it.


Oh look, lots of adorable little sloth-parasites are hatching.


I’ll see your HP Lovecraft dreams, and raise you a Langford Parrot.


It’s duplicate posts all the way down…


A few years ago, I asked “What happens when random data is decompressed?” The question has just been answered.


According to this article, this type of neural net, called a “Boltzmann machine”, actually has to go through something like sleep/wake cycles to function correctly, so there’s speculation that real brains may work in an analogous way and that this could explain why sleep is universal in creatures with brains. From the article:

The synapses in the network start out with a random distribution of weights, and the weights are gradually tweaked according to a remarkably simple procedure: The neural firing pattern generated while the machine is being fed data (such as images or sounds) is compared with random firing activity that occurs while the input is turned off.

Over the past five to 10 years, studies of brain activity during sleep have provided some of the first direct evidence that the brain employs a Boltzmann-like learning algorithm in order to integrate new information and memories into its structure. Neuroscientists have long known that sleep plays an important role in memory consolidation, helping to integrate newly learned information. In 1995, Hinton and colleagues proposed that sleep serves the same function as the baseline component of the algorithm, the rate of neural activity in the absence of input.

The easiest way for the brain to run the Boltzmann algorithm, he said, is to switch from beefing synapses up during the day to whittling them down during the night. Giulio Tononi, head of the Center for Sleep and Consciousness at the University of Wisconsin-Madison, has found that gene expression inside synapses changes in a way that supports this hypothesis: Genes involved in synaptic growth are more active during the day, and those involved in synaptic pruning are more active during sleep.


Has anyone considered that these neural networks might be conscious in some way while doing this stuff, since they apparently work similarly to real brains? I know it sounds like a weird idea, but I just thought I’d throw the idea out there.


Can anyone come up with a reasonable definition of ‘conscious’ for which these systems are not obviously conscious?


I’d like to see someone comparing these images with what patients with hallucinations of any kind describe. I bet there are similar mechanisms at work.


I think instead of conciousness we’re seeing something more like a “Platonic form” but completely naked, uncontexted, in a pure graphic representation that might strike a person as a bit rude. It’s like if a chart were drawn that showed how fingers must be tensed while doing a pull-up, we’re seeing how a neural network “grips” a recognizable form, which produces images with an uncanny likeness to dreams.

What’s more exciting to me than re-creating consciousness is the possiblity of mathematically disproving our own.



This topic was automatically closed after 5 days. New replies are no longer allowed.