Philippe Boisnard
Online Projects
- ADN-MOVIE Software for performing film analysis with advanced AI tools. (2015-2026)
- Latent Flora — Structural Unconscious 1,100 AI-generated flowers, SAM segmentation, UMAP latent space walk (2023-2025)
- Latent Art History — Computational Hermeneutics 1,000 artworks analyzed through CLIP semantic axes and SAM segmentation (2026)
- Art Explorer — Visual Search Zone of Attention - AN INTROSPECTION INTO THE SCOPIC REGIMES OF CLIP VIT-B/16 (2026)
- Temporal Sedimentation Benchmark Interactive CLIP tool for visual temporal exploration (2026)
Writings on AI & Art
- On the Structural Unconscious of Generative Models 2026 — Examining latent biases in image generation
- From Metrology to Hermeneutics: Beyond Manovich 2026 — CLIP as a tool for computational art criticism
- Segmentation as Interpretation 2025 — SAM and the cultural bias of computational perception
- Prolegomena to a Post-Aesthetics of Artificial Imaginations 2024 — The acceleration of the use of generative artificial intelligences (AI), since 2015 and the turning point operated by Deepdream, tends to obscure a real analysis of what could be defined as artificial imagination. AIs are either reduced to simple instruments or thought of according to a form of techno-theologism. Our research tends to suspend any form of judgment in order to phenomenally grasp the emergence of these AIs. By taking up the question of Hegel's aesthetics and of art as the free production of the mind, but by moving it towards the question of generative AIs and therefore of a post-aesthetics, this article will show the phenomenal specificity of images generated by AI.
Practice & Research
Philippe Boisnard is an artist, writer, and researcher working at the intersection of artificial intelligence, digital poetics, and computational aesthetics. His practice explores the latent structures embedded within generative models — the unconscious patterns, cultural biases, and perceptual habits that machines inherit from their training data.
His current work focuses on using segmentation models (SAM) and vision-language models (CLIP) as instruments of critical analysis, treating AI not as a creative tool but as a mirror that reveals the hidden architectures of visual culture. By walking through the latent spaces of generative and analytical models, his projects make visible what these systems have learned to see — and what they systematically fail to perceive.
His approach draws on a lineage that includes Lev Manovich's Cultural Analytics, extending computational art analysis from physical measurement to semantic interpretation. Where earlier methods measured brightness and saturation, Boisnard's work asks whether a painting is sacred or profane, ordered or chaotic — questions that only became computationally tractable with the emergence of multimodal AI models.