Wenqian GAO: Chameleon

 
  • ©,

  • ©,


Artist Statement:


    Summary

    Chameleon change the color of skin according to the surrounding environment for hiding the body. This is a pixel chameleon generated by AI. Artificial intelligence one day can be like a chameleon hide himself with pixel camouflage? How will be the relation between Artificial intelligence and the nature?

    Abstract

    Chameleon change the color of skin according to the surrounding environment for hiding the body, can not only avoid predators, and also confuse their prey. I use against neural network, I input a lot of images of chameleon various parts to the machine learning algorithm, the neural network generated the color-changing chameleon skin based on these pictures. Artificial-intelligence one day can be like a chameleon hide himself with pixel camouflage? Work is divided into two versions, image and image + interaction: in the interactive version, the color of the chameleon can change color according to the color of the catch by the electron trap.


Technical Information:


    Machine learning style-gan, video on screen of led (or projection). Work is divided into two versions, image and image + interaction: in the interactive version, the color of the chameleon can change color according to the color of the catch by the electron trap.


Process Information:


    The biggest challenge is that I didn’t know how to use the adversarial neural network to generate these images at first. I installed many python modules on my computer to build the neural network and learn the relevant machine learning knowledge. I spent a lot of time to find the tools of artificial intelligence on the Internet. Finally, I  found the most convenient and best way on RunwayML.

    Another difficulty is about the collection of chameleon images. I tried to find a chameleon, but it seems illegal to breed chameleons in China. I haven’t found it for various reasons. Later I thought why I didn’t use the pictures on the Internet? The pictures on the Internet sometimes It is difficult to distinguish the authenticity, and it is as deceptive as the skin of a chameleon.

     


Other Information:


    Inspiration Behind the Project

    This work originated from my thinking about the digital world and the real world of humans. In my past works, I often think about the skin of the digital world presented to the real world through pixels: such as textures in 3D modeling, “pixel camouflage”, the Painted skin of artificial intelligence. I hope to continue to explore this theme in this work. I often think about whether there will be an interactive evolution between humans and artificial intelligence?  For example, the evolution of nature was affected by natural selection and artificial selection and will there be “artificial intelligence selection” in the future? At the beginning, I think that artificial intelligence first learns to be like natural creatures, learn to protect itself. I want imagine a skin for artificial intelligence, so I thought of the skin of a chameleon.  And the Generative Adversative Nets itself is just a game of evolutionary algorithm Therefore, the content and technical form of the work fit together. 

    Key Takeaways for the Audience

    While humans are training artificial intelligence, artificial intelligence also trains humans. Humans and artificial intelligence are evolving together. We need to find a way to coexist harmoniously with “artificial intelligence-human extension.”


Website: