Material Intelligence
Scale 1: The Substrate
We move beyond simulating intelligence in silicon. Our mission is to forge it directly from physics. We architect matter that learns, self-corrects, and evolves—a true physical basis for cognition not designed by humans, but discovered by the material itself.
From chaotic particles to a complex, emergent pattern.
Foundational Pillars of Material Intelligence
To achieve MBI, a system must synergistically integrate four key capabilities:
Embodied Perception
Systems must integrate multiple sensory inputs not as discrete data points, but as perturbations to a continuous physical field.
Intrinsic Computation
We leverage nonlinear dynamics and phase transitions. The physics itself becomes the algorithm.
Active Memory & Learning
Information is encoded as physical state (polymer conformations, defect patterns). This memory actively shapes future responses.
Self-Organization
Designing systems with rules for self-correction and repair, allowing complex architectures to emerge and persist.
The Generative Engine: Self-Organization
Intelligence begins when order spontaneously emerges from chaos. Our systems are not constructed; they grow. Driven by a continuous flow of energy, they harness local interactions to self-assemble into complex, functional architectures without an external blueprint, forming the very foundation upon which all other cognitive functions are built.
Disordered particles self-organizing into a complex structure.
The Embodied Mind: Active Memory & Physical Learning
Matter that cannot learn is not intelligent. Our substrates physically change based on their history. Memories are not stored as abstract bits, but are etched into the material's very structure—polymer conformations, defect patterns, and energy landscapes. The material becomes a physical model of its experience.
A particle explores a potential landscape, physically carving a more efficient path (memory).
The Drive for Persistence: Self-Correction & Repair
Robustness is a hallmark of life. We design materials with an intrinsic drive toward integrity. By physically encoding a 'target morphology' as a low-energy state, these systems can autonomously detect damage, harness energy, and trigger processes to heal themselves, ensuring persistence against entropy.
A 2D material autonomously detects and repairs a complex fracture.
The Future: Self-Replication with Variation
Beyond repair lies replication: the seed of evolution. The challenge is to architect systems that can attract ambient building blocks to construct copies of themselves. Crucially, this process allows for slight variations, creating a population of diverse agents upon which selection pressures can act, enabling a truly material-based evolution.
An original structure (left) assembles a copy of itself (right) from ambient particles.
The Behavior: Emergent, Non-Human Intelligence
We do not program solutions; we cultivate the conditions for them to emerge. By leveraging criticality and complex dynamics, our systems discover solutions and behaviors that were not explicitly designed—a truly alien form of non-biological cognition that computes by physically exploring the possibilities of its world.
Simple, local rules in a grid create complex, unpredictable global patterns.
Architect the Future of Intelligence with us
Creating material intelligence is a grand challenge that requires a collective of physicists, chemists, biologists, and AI researchers. Our ecosystem is designed to merge these diverse expertises into coherent, mission-driven teams. Our institute fosters this by orchestrating collaboration and building upon foundational work, such as that presented at the Intelligent Soft Matter Workshop.
Our Launchpad is an intelligence design tool. UseDiscovery Engine to find collaborators, combine nodes of expertise, search for labs with critical equipment.