David Baker, winner of the Nobel Prize for Protein Design, now uses AI to create molecules not found in nature, pushing the boundaries of biological engineering beyond what evolution has produced. His work at the University of Washington's Institute for Protein Design signals a fundamental shift from describing natural proteins to inventing entirely new ones.

What You Need to Know

Proteins drive nearly every biological process, and designing them from scratch with AI opens the door to custom enzymes, drugs and materials. Baker's lab uses deep learning to predict protein shapes and functions without relying on natural templates. This approach could accelerate drug discovery, create new biomaterials and raise safety questions about synthetic organisms.

A New Era for Protein Engineering

For decades, scientists studied proteins as they exist in nature. Baker reversed that model. Instead of analyzing existing structures, his team builds proteins that have never existed, starting with a desired function and working backward to find the amino acid sequence. That process, once slow and laborious, now runs on AI models that generate thousands of candidate designs in hours.

The Nobel Prize in Chemistry awarded to Baker in 2024 recognized his foundational work in computational protein design. But the field has moved far beyond the early Rosetta software. Modern neural networks, trained on vast protein structure databases, can hallucinate novel folds and active sites.

Now He Uses AI

Baker's current research relies heavily on generative AI. His lab has developed tools that treat protein design like language generation. The AI learns the grammar of protein sequences and produces candidates that are stable, soluble and capable of specific biochemical tasks. These molecules are not variations on natural proteins. They are entirely new architectures.

Several breakthroughs have emerged from this pipeline:

  • Custom enzymes: AI-designed catalysts that perform reactions not found in any known organism.
  • Self-assembling nanomaterials: Protein cages and lattices that could deliver drugs or form microscopic devices.
  • Therapeutic proteins: Molecules designed to bind specific disease targets with high precision.

Create Molecules Not Found

The ability to create molecules not found in nature carries both promise and risk. On the practical side, synthetic proteins could replace toxic industrial catalysts, enable new vaccines and produce sustainable materials. But the same tools could be used to design harmful proteins or enable bioweapons. Baker has been vocal about the need for responsible development and has released many designs openly to foster collaboration and oversight.

Regulatory bodies face a challenge. Current biosafety frameworks were built around natural organisms and known pathogens. AI-generated proteins with no natural counterparts may not fit within existing guidelines. The scientific community is still debating how to assess risk when the sequence has never existed in nature.

Why This Matters

For the pharmaceutical and materials industries, the implications are immediate. Drug discovery cycles could shrink from years to weeks. Custom biomaterials could replace petroleum-based plastics or enable new forms of tissue engineering. For regulators and policymakers, the speed of innovation demands updated safety protocols. For the public, the arrival of AI-designed biology means that the molecular world is no longer limited by natural evolution. Nature David Baker and his colleagues are rewriting the rules of what is possible, and society must adapt as quickly as the algorithms advance.

The convergence of Nobel Prize-winning protein science with generative AI has created a new branch of engineering. Biology is no longer something humans study. It is something humans design.