When the visitor enters the gallery space a big eye on a screen placed over a security camera indicates that someone is watching. When the person moves forward a screen indicates that different body parts of the person are detected also the face. In the end of the hallway(stairs) the visitor can observe several portraits changing on screens. These portraits are generated using a GAN (Generative Adversarial Network) that is trained on faces. Each visitors face is captured and added to the training set. Then new faces are generated and displayed on the screens. Finally the last screen has a black and white image that: "represents the ultimate identity of the AI ghost, which is based on the most unique of all previously generated faces."(1)
(1) https://www.howeb.org/portfolio-item/what-a-ghost-dreams-of-2019/