Researching intelligence
that begins with seeing.
INsVI studies what language-centered artificial intelligence leaves unresolved: spatial relations, composition, materiality, ambiguity, visual judgment, and knowledge carried by images themselves.
Visual intelligence is more than recognition.
Naming objects is useful. It is not the same as understanding how an image is organized, what relationships it establishes, or why its visual form matters.
Perception and representation
Study non-verbal visual representations that preserve spatial, compositional, temporal, and perceptual information.
Art as a test environment
Use fine art to investigate ambiguity, interpretation, imagination, visual structure, and culturally situated perception.
Data with depth
Develop approaches to visual datasets that value provenance, context, process, expert judgment, and ethical documentation.
Six connected lines of inquiry.
Spatial and relational reasoning
Depth, occlusion, viewpoint, scale, orientation, object permanence, and physical causality.
Visual form and composition
Hierarchy, rhythm, proportion, balance, tension, color relationships, and image structure.
Visual datasets and provenance
Smaller, deliberately curated collections with traceable sources and meaningful context.
Interdisciplinary by necessity.
Visual intelligence cannot be addressed by computer science alone. INsVI connects fine arts, computer vision, data science, design, philosophy, cognitive science, and AI ethics.
Our work is organized as active inquiry: defining research questions, assembling visual evidence, designing experiments, documenting methods, and publishing findings as they mature.