This image is a computer’s take on a mushroom!
This project was created by Joel Lehman, an assistant professor at the IT University of Copenhagen. The work is untitled, but it was created in 2016. Link: https://www.popsci.com/creative-ai-learns-to-sculpt-3d-printable-objects.
I really liked this project because I thought it did a good job of incorporating a human touch to a rigid process. In fact, the algorithm mimicked the human learning process via deep neural networks.
From what I can tell through the use of image recognition, the computer can get fairly close to something that resembles an object. Though not perfect, this is something unique. The computer is in essence drawing an object in its own style. In a way the algorithm is generating objects at random and through many many repetitions is able to produce something close to the object it has been trying to draw.
The artistic sensibilities are manifested in the creation algorithm. While the algorithm is random, the process in which it is generated (better versions serve as starting points for future generations) helps to determine the style of the content being produced. In fact, one can tell that all the final images were created by the same algorithm.