One of Google’s new projects, known as Inceptionism, gave an artificial neural network the task of designing a visually stimulating piece of art. Artificial neural networks are designed to mimic biological neural networks–specifically brains. When you hear about facial recognition, speech recognition or artificial consciousness, it’s pretty much the same thing.
We’re giving computers the ability to think for themselves faster than you might expect, and one aspect of artificial intelligence is creation. What did Google’s computer create? Check it out:
How did Inceptionism come to be?
Before assigning the computer its task, Google researchers trained the program to recognize a variety of shapes, objects, and biological creatures. They did this by feeding the system millions of photographs. You can imagine the magnitude of teaching a computer to differentiate these samples from one another just like humans do. Many similar projects have made enormous progress towards this goal.
Each layer of Google’s artificial neural network has a different job. One layer might identify and categorize edges, lines and other basic shapes while another might interpret the various data collected to form the most accurate identification possible.
Eventually, the researchers behind Inceptionism decided that the neural network was also capable of generating an image of its own based on a foundation image. They provided it with a photograph of antelopes, and it came up with this:
If you think it looks like the computer is on drugs, you’re not alone. In another experiment, the researchers fed the neural network photographs of clouds, inspiring a comparison game we’ve all played:
“What does that cloud look like?”
“It looks like a baby Dumbledore!”
One of the researchers explained the purpose of the task:
If a cloud looks a little bit like a bird, the network will make it look more like a bird. This in turn will make the network recognize the bird even more strongly on the next pass and so forth, until a highly detailed bird appears, seemingly out of nowhere.
Here are a few examples of what Inceptionism came up with:
If you’d like to see a few more, then watch the following short video. Otherwise, tell us what you think of Google’s Inceptionism research! Are artificial neural networks and artificial intelligence becoming too powerful?
2 Comments
Sign me up to your blog pease and thanks for the great work!
Hi Darred,
Thanks for the comment! If you go back to our homepage, scroll to the very bottom and you’ll find where you can sign up for our newsletter.
Sincerely,
Tech Gen Mag