Google’s DeepDream AI can be used to make some wild art. The artificial intelligence can try to match what it sees in videos or still images with images in its huge database, then can overlay what it finds over the original. The results are strange, and some would say nightmarish, collages. It was originally trained on images of dogs, so it imagines a lot of dog faces.

Here are some examples:

I think it’s safe to say that anyone who had the chance to watch The Joy of Painting with Bob Ross absolutely loved the show. If you didn’t fall asleep to the soothing sounds of his voice first, you usually got to witness some pretty awesome painting by the man with the best afro in the entire world. However, in all your time since watching the show and having fond memories of the presence of Ross, I seriously doubt you ever though of Artificial Intelligence. And I highly doubt you thought of LSD. I bring this up because….


This video shows some of the unreasonable effectiveness and strange inner workings of deep learning systems. The unique characteristics of the human voice is learned and generated as well as hallucinations of a system trying to find images which are not there.

Link

Ever since Google released the neural network DeepDream in 2015, we’ve learned that AI is capable of creating things that are far trippier (and far more alarming) than anything the human mind can dream up. Filtering videos like scenes from Fear and Loathing in Las Vegas (1998) through a neural network can have some truly disturbing results, as the computer reaches to find meaning and shapes where there aren’t any. The results are a colorful collage of mashing limbs, teeth, and animals that rotate and transform as the video rolls. In other words, the worst thing we’ve ever seen.

Link

More on YouTube:

If you are a bit technical, here’s how to install/use DeepDream yourself

Here’s a link to a community using DeepDream for various things.