Word on the street is artificial intelligence is out to steal your job, but we often forget about all the good things it also does for us: Like helping us draw penises.
One industrious researcher took it upon themselves to feed a recurrent neural network with 10,000 dick doodles in order to teach it how to draw the human phallus — and the result is equally glorious and abhorrent.
Dick RNN is like any other web-based sketching board, with one simple exception — it will take any shape you draw on it, and try to turn it into a pecker.
The neural network was fed a healthy diet of 10,000 tallywhacker doodles to acquire this impressive skill. Having gotten intimately familiar with a wide variety of dongs, Dick RNN will take your input and keep drawing unique penises until you make it stop.
When you spot a schlong you dig, you can click the PNG (or SVG) button at the upper right corner to save the sketch.
Is it perfect? Erm, I’ll let this penis doodle speak for that:
You’ll certainly find some phallic shapes you probably haven’t seen before, but for the most part — it gets things right.
I noticed that starting with the balls is the easiest way to help the AI to draw an authentic penis. Pro-tip: Don’t start with the tip — unless you want to witness some absolute monstrosities. On second thought, maybe try that, it’s pretty funny.
Dick RNN is based on Google’s Sketch RNN, a neural network trained to generate coherent sketch drawings. In fact, one of the possible uses its creators David Ha and Douglas Eck imagined was to build an app that takes “crude, poorly sketched drawing[s] and generate[s] more aesthetically looking reproductions.”
Well, it seems someone took their suggestion into consideration.
You can check out Dick RNN by clicking here. You should also have a gander at this alternative implementation that draws multiple dicks at the same time. Ah, and if you wanna have a closer look at the code behind this powerful technology, head to this GitHub repository.
Now go make your parents proud.
Published April 24, 2020 — 13:01 UTC