Maja Petric & Mihai Jalobeanu
Environment

Maja Petric & Mihai Jalobeanu

The Glitched Sublime: Arctic Poppies

Visuomotor Neuro-symbolic AI Robotics Neuro-symbolic AI Embodied AI Interactive

The Glitched Sublime is a series of autonomous, data-driven lightbox sculptures developed by Maja Petric in collaboration with roboticist and AI scientist Mihai Jalobeanu. Each wall-mounted sculpture functions as a climate artifact, translating live Arctic environmental instability into physical form through robotics, artificial intelligence, light, and layered optical systems.

The sculptures extend five inches into space, combining illuminated depth, fabricated Arctic poppies, and lenticular optical surfaces that create shifting atmospheres of shimmer and parallax. The flowers appear suspended within weather rather than behind glass, hovering between preservation and disappearance.

The sculpture is built by an autonomous robotic system, in response to live Arctic climate data. Whenever the system detects a new threshold of ecological stress, such as temperature anomalies or destabilizing climatic deviations beyond historical norms, another fabricated flower is added to the sculpture. Constructed from resin, paper, and synthetic materials, the flowers suggest a machine attempting to reconstruct nature through entirely artificial means, as though trying to remember a living ecosystem it never directly experienced.

To translate invisible climate data into an evolving physical composition, the installation relies on an autonomous AI-driven decision-making system. The robotic arm functions only as the mechanism of execution; the central intelligence of the work exists within the AI and software determining where every flower belongs. Rather than following pre-programmed coordinates, the system continuously evaluates the existing composition and generates new spatial decisions in real time.

This process is implemented as a neuro-symbolic pipeline. A large-scale foundational model establishes the process and structural logic of the artwork: interpreting climate data, understanding spatial constraints, and generating possible compositional arrangements. A second, custom AI model acts as a critic, evaluating these possibilities according to a highly specific aesthetic language developed from the artist’s own compositional intuition. Through simulated generations, pairwise comparisons, and iterative training, the system learned to recognize tensions between density and absence, order and instability, coherence and collapse.

During live operation, whenever a climate anomaly triggers a new intervention, the system generates and evaluates multiple spatial configurations across a topological grid. The critic model continuously assesses these proposals in real time, selecting the configuration most aligned with the evolving aesthetic state of the work, what the artist describes as the “glitched sublime.” These abstract spatial decisions are then translated into visuomotor primitives that control the robotic arm’s kinematic movements, allowing the system to physically pick, maneuver, and insert each fabricated flower into the sculpture.

As ecological instability intensifies, the sculptures become increasingly dense and visually radiant. Paradoxically, the most luminous and abundant compositions correspond to the highest levels of environmental disruption. Each sculpture evolves uniquely over time, with its final form determined directly by the climatic conditions recorded throughout its existence.

maja-petric
maja-petric
maja-petric
maja-petric
M
About The Artists

MAJA PETRIC

Maja Petric is an artist working with light, optics, AI, robotics, and real-time environmental data to create immersive installations and responsive systems that transform ecological change into embodied sensory experience. For more than two decades, she has collaborated with computer scientists and engineers to develop custom technologies that redirect systems designed for optimisation and monitoring toward perception, memory, and emotional connection with the natural world.

Her work has been exhibited internationally at Phillips Auction London and Hong Kong, FOR-SITE, Henry Art Gallery, and MadArt Studio, and commissioned by National Geographic, Meta, Microsoft, Google, and Amazon. Petrić is the recipient of the Digital Art Award for Innovation, the Lumen Prize for Art and Technology, and the Richard Kelly Light Art Award. Her work has been featured in The New York Times, BBC, Fast Company, and Vogue, while her collaborative technological systems have been presented at NeurIPS, ICCV, and ISEA.

Petrić holds a Ph.D. in Digital Arts and Experimental Media from the University of Washington and a Master's degree from NYU Tisch School of the Arts.

MAJAPETRIC.COM


MIHAI JALOBEANU

Mihai Jalobeanu is a computer scientist and robotics researcher specialising in artificial intelligence, machine learning, computer vision, and embodied robotic systems. During eleven years at Microsoft Research, he co-founded the Robot Learning Group and contributed to foundational research in adaptive robotics and intelligent systems. For more than a decade, Jalobeanu has collaborated with Maja Petrić to develop the custom AI and robotic systems at the core of her installations. Their technologies are built from the ground up, allowing the conceptual and sensory demands of the artwork to shape the technology itself rather than adapting existing commercial systems. Their collaborative work has been presented at NeurIPS, ICCV, and ISEA.

DEXMAN.AI