This AI model enables the blending of two sets of images. For the first set, I selected 1,000 AI-generated faces that do not exist in reality, created by synthesizing features from real people. The second set consists of my own images, with the aim of seeing how the system identifies and incorporates elements of my face into the results. During the process, I observed several intriguing phenomena in the progressively generated fakes. 

The most fascinating and personally compelling aspect was watching how facial characteristics from the original dataset were extracted and subtly "stitched" onto my own face. For instance, an elderly person's wrinkles might transfer onto my face, sometimes blending in unexpected ways—such as being mistaken for the contours of my mouth. This often resulted in surreal effects, where it looked like two faces merged within one head. It reminded me of Francis Bacon's paintings, where blurred and distorted portraits evoke a sense of unease and psychological instability. 

Through this experiment, I aim to highlight the issues of information overload and emotional manipulation brought about by the internet. Every day, we are exposed to thousands of pieces of content created by individuals. In the battle for attention in this era of information excess, content creators are compelled to manipulate our emotions to achieve their goals. One post might evoke sadness, while the next might provoke anger. Empathy, a human nature, has been overexploited and abusively used in the digital age, eventually leading to a kind of imagined fusion of subjectivity.

Before the performance, I was skeptical about the outcome due to my critical stance on AI. I doubted that AI could seamlessly merge my face with someone else's in a way that would appear convincingly real. To test its limits, I fed the AI model my most exaggerated and absurd facial expressions. As expected, the result was even more distorted and surreal than my original inputs. -John

John Wang is currently studying Interactive Media Arts at NYU.