SAID Lab · est. 2026

Engineering AI forthe future of science.

We exist at the intersection of physical reality and machine intelligence. SAID Lab is a scientific research laboratory dedicated to advancing humanity's understanding of nature itself. We build models that obey the laws of physics, that can be read, and that collaborate at machine speed. We believe the next scientific revolution will emerge from collaboration between humans and Scientific AI.

Definition of Scientific AI01 / 02
i.

Physical Grounding

Models constrained by the laws of nature. Differential equations, conservation laws, and thermodynamic principles embedded directly into model architectures. Scientific validity prioritized over statistical plausibility.

ii.

Interpretability

Black-box systems are insufficient for scientific discovery. Symbolic regression, attention mapping, and human-readable hypotheses. Researchers should understand not only what the AI predicts, but why.

iii.

Autonomous Collaboration

Multi-agent systems we call digital scientists. They generate hypotheses, simulate experiments, review findings, critique one another, and iterate, at machine speed, while remaining grounded and interpretable.

Eight pillars, one method.

01 · Pillar

Scientific AI for Biology

Protein modeling, systems biology, drug discovery, genomics, and cellular simulation. Physics-grounded models for molecules whose function is mathematically constrained, not merely statistically likely.

Read more →
02 · Pillar

Scientific AI for Neuroscience

Brain modeling, cognitive systems, psychiatric computational models, and neural representation learning. Closing the loop from neural data to mechanistic theory.

Read more →
03 · Pillar

Scientific AI for Quantum Computing

Quantum simulation, algorithm optimization, AI-assisted system discovery, and architecture exploration. Treating decoherence as an optimization problem.

Read more →
04 · Pillar

Scientific AI for Physics

Theoretical physics, simulation systems, dynamical systems, and scientific modeling. Differential-equation-constrained networks and symbolic regression for the recovery of laws.

Read more →
05 · Pillar

Scientific AI for Mathematics

Automated theorem exploration, symbolic reasoning, neural-symbolic systems, and mathematical pattern discovery.

Read more →
06 · Pillar

Scientific AI for Machine Learning

Treating ML itself as a scientific object. Interpretability, physically grounded learning, multi-agent architectures, and the limitations of foundation models.

Read more →
07 · Pillar

Scientific AI for Economics

Complex systems modeling, economic simulation, emergent behavior analysis, and policy modeling. Economics as a quantitative science.

Read more →
08 · Pillar

Scientific AI for Chemistry

Molecular simulation, reaction prediction, material discovery, and computational chemistry grounded in conservation laws.

Read more →
All pillars

See the full research index

Every domain, every preprint, every weight. Open.

Explore research ↗
Manifesto02 / 02

AI is being pointed at the wrong problems.

Humanity has created the most powerful computational tools in history. Most of that capacity is pointed at advertising, content, and B2B SaaS.

Meanwhile, neuroscience remains poorly understood, scientific progress is slowing in many fields, experimental costs are rising, and researchers cannot keep up with the literature. SAID Lab believes artificial intelligence should primarily be used to accelerate humanity's understanding of physical reality.

We optimize for scientific discovery per year, not revenue per quarter.

We hire bilingual researchers, people fluent in both a science and computer science. We publish openly and reduce barriers to participation.

We are a scientific infrastructure institution for the 21st century. We view scientific AI infrastructure the way the world views observatories, public libraries, and open datasets: as a public good.

Read the full manifesto ↗

Let's discover together.

We respond within a week. Send research notes, papers in progress, or a paragraph about what you're circling. We welcome collaborators, advisors, and supporters who share our long horizon.