Pichardo
United States
Artist Statement
Description
Machine Gaze is a generative self-portrait built from the detritus of my online life: every post, search, and fleeting “like” accumulated since 2005. A small embedded computer continually sifts through this personal archive and, using a custom neural network, continuously reconstructs my face as a fluid, shifting image. The portrait never fully settles—it flickers between recognition and abstraction, revealing the biases and distortions inherent in algorithmic vision. In this way, the work mirrors how social media platforms transform our identities into probabilities and market segments, highlighting the often unseen ways technology reshapes our sense of self.
Process
Social platforms promised connection, but after twenty-plus years they operate mainly as data-collection engines. Every status update, selfie, and search query is logged, sorted, and priced. Over time those logs start to define us more than our own memories do. Deleuze calls this condition the “dividual”: a person reduced to interchangeable data points. Machine Gaze grew out of my need to see what those data points add up to. I gathered my own posts, profile images, and location traces and fed them into a model that rebuilds my face in real time. By showing that reconstructed image in the gallery, I’m inviting viewers to consider two simple questions: How accurately does the system “know” me, and who benefits from that knowledge? The project isn’t about condemning technology; it’s about making the hidden perspective of the platforms visible, so we can decide what kind of mirrors we want to keep using.
Tools
Machine Gaze was an iterative process that started with exporting archives of my data from Google, Facebook, and Instagram. For each image where I was in the frame, I used Mediapipe's face landmark detection model to cut out my eyes, nose, ears, mouth into separate images. I wrote a collage generation script that would take those images and randomly combine them into all possible permutations of collages of my face. These collages would become the training data for my face generation model. I tried multiple approaches, including diffusion models and GANs. However, I settled on CPPNs (compositional patter-producing networks) because I was drawn to the smooth organic gradients and theoretically infinite resolution so I developed a custom GAN with a CPPN as the generator. This simple network learns to map pixel coordinates and a latent value to a pixel color (x,y,z) -> (r,g,b), in this regard it is extremely simple and that is the beauty. Since it's just a multi-layer perceptron, it was possible to transpile the model and weights to a fragment shader which is able to do inference in real-time on the GPU on a low-power device like a Latte Panda, which continually paints morphing, animated, non-repeating, smeared, portraits of my face. It's all packaged in a square aspect ratio LCD screen inside a wooden picture frame.
As a creative entrepreneur, Liri loves to envision, execute, and push new ideas and bring them to life. Liri is Co-Founder of OFFF TLV the global festival of design, and serves as Director of Brand at Guesty.
Who are wein the eyes of the modern world? Or more precisely—are we merely a reflection of whatthe machine, trained to observe us, thinks we are? It learns from our searches,our likes and dislikes, building a version of us that feels oddly familiar, yetnot quite human. Machine Gaze echoes this philosophical tension.
Thisartwork offers a powerful and unsettling response: a portrait in motion. Andyet, the outcome feels like a distorted glitch—haunting, incomplete.
Artificialintelligence is generating a new kind of human perception—one that iscollected, categorized, and analyzed by algorithms. The result may betechnically accurate, but is often distant and deformed. We are at thebeginning of an era in which, I believe, AI will increasingly be used to createthings not for us, but for the machine’s perception of us. This raises afundamental question: will this technology truly serve humanity, or merelyserve the machine’s idea of humanity
The artworkcaught my eye because I recognized the semblance of a human face. But it wasthe distortion that drew me in—I found myself staring at it. It felt like looking at a sick person: sadeyes, heavy with a story.
Is this us
Or betteryet—is this me?
