Experts, including Jonathan Losos and Zachary Blount, support the findings, emphasizing that evolution could have taken many different paths that were not realized in Earth's history.

ESM3 operates by modeling protein sequences, structures, and functions through discrete tokens, marking a significant departure from previous methods that focused primarily on protein sequences.

The EvolutionaryScale Model 3 (ESM3), developed by former Meta researchers, is an advanced AI system designed to generate proteins based on extensive biological data.

This AI model utilizes a massive training dataset comprising 771 billion unique tokens from 3.15 billion protein sequences, 236 million protein structures, and 539 million functional protein annotations, and it can manage up to 98 billion parameters.

Currently in public beta, ESM3 is accessible via an API, enabling scientists to programmatically engineer proteins or use interactive applications through a web browser.

In a groundbreaking experiment, ESM3 was tasked with creating a new green fluorescent protein (GFP) and successfully produced a novel protein named esmGFP, which exhibits only 58% similarity to existing GFPs, effectively simulating 500 million years of evolutionary processes.

This innovative approach not only enhances the understanding of naturally evolved proteins but also facilitates the creation of new proteins for various applications in medicine and environmental remediation.

The ESM3 experiment highlights the concepts of contingency and determinism in evolution, suggesting that while certain evolutionary paths were possible, others were not realized due to historical circumstances.

A recent study published in the journal Science utilized ESM3 to identify alternative pathways for protein design, which could have significant implications for developing new therapies.

The genetic sequence of the newly developed fluorescent protein is notably different from existing known fluorescent proteins, representing a major breakthrough in protein engineering.

This exploration raises intriguing questions about whether human evolution was an inevitable outcome or a series of random accidents, further fueling debates on the nature of evolutionary pathways.

Researchers can access the EvolutionaryScale Forge API through a free academic access tier, or utilize the open model's code and weights for their projects.