Art via Artificial Intelligence

At the end of August a new paper was posted that showed how to “teach” an artificial neural network (ANN) the style of a given painting or piece of art.  It built off of earlier work by some people at Google they called inceptionism, from just a few months earlier.

Don’t worry.  The machines aren’t coming for your paint brushes.  Despite their name, ANNs are just a class of algorithms that non-linearly transform a linear combination of signals, over and over again.  These algorithms are modeled off of how your brain works, and have come a long way in the last 10 years due to new implementations called Deep Learning.  You should check it out.

While all this technical progress has been great (voice recognition for Siri and Google Voice Search doesn’t suck anymore, Facebook can auto-tag your face), the benefits have been mostly marginal.  Soon though, cars will be driving themselves.  The 10% of the economy that is transportation will be completely disrupted.  Then things will happen more quickly.  We need to be thinking about what a post-work/post-scarcity economy looks like.  In the meantime…

Someone has turned the Neural Algorithm for Artistic Style into a service!  DeepArt.io  You upload a picture you want reinterpreted by an algorithm (me and Bernie), upload a painting style you want emulated (MC Escher’s Smaller Smaller), and hit a button!  If you pay $2 you get it in minutes.  If you don’t pay… I haven’t not paid yet, so I don’t know how long it takes, but it takes less time than it says it will.

It’s the ultimate Instagram filter.  Enjoy!

MeBernieEscher1

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s