In 2018 I created a series of experiments titled "Neural Glitch" in which I deliberately manipulated the inner weights of custom trained GANs by permuting and altering them.

As a result these models that had originally been tasked to enhance the inputs they were given started to misinterpret those inputs both on an aesthetic and a semantic level.

The artificially introduced errors which emphasise the convolutionalnature of these early models give us a glimpse into the inner workings of these complex networks and at the same time result in an aesthetic which is almost unique to a certain short phase in the rapid development of AI art.