Former AI Researchers Raise $40M to Develop Biological AI Models
Former researchers from Meta have launched a new startup called EvolutionaryScale, raising $40 million in funding. Led by Alexander Rives, the team aims to develop AI language models (LLMs) for biology. These models will be used to predict the structures of proteins, which are crucial for developing drugs and other important molecules. The startup plans to dramatically scale up the size of its AI models to advance its research efforts. Lux Capital led the funding round, with prominent AI investors Nat Friedman and Daniel Gross also participating. EvolutionaryScale aims to contribute to the field of biology by developing models that can help cure diseases, clean up pollution, and manufacture industrial chemicals.
Predicting protein structures is a complex task due to the interactions between thousands of atoms that govern their shape. However, accurately predicting these structures is crucial for designing drugs that can properly bind to them. DeepMind’s groundbreaking AlphaFold system, released in 2020, made significant strides in this field. Nevertheless, there is still much work to be done in understanding how drugs interact with the vast number of proteins. EvolutionaryScale’s model, as highlighted in a paper published in Science, has shown promise by making predictions 60 times faster than AlphaFold.
Although AI has brought incremental improvements to drug development efficiency, there has yet to be a major breakthrough in the biology field. The team from Meta left the company in April, as Meta shifted its focus to commercial efforts. The marketplace for AI in biology is currently small, with traditional pharmaceutical firms skeptical about its potential. However, there are companies like Schrodinger that continue to sell products based on older molecular modeling methods. Despite this skepticism, there has been a surge of interest in transformer-based AI research, with companies like Inflection AI, Cohere, Adept, and now EvolutionaryScale raising substantial funding.
The development of AI models for protein folding and other biological tasks requires significant investment. Companies like DeepMind, Insitro, and Recursion have raised billions of dollars to advance their research in this field. However, the average time from drug discovery to FDA approval is still around 7-10 years. EvolutionaryScale acknowledges the long road ahead, projecting high costs for its compute power and emphasizing that it could take a decade for its models to have a tangible impact on product and therapy development.
Despite the challenges, EvolutionaryScale is determined to make a breakthrough in AI for biology. The company plans to scale its AI model and increase its size, aiming to achieve a “capability breakthrough” similar to the advancements seen in natural language processing in 2018. The startup envisions building a new model each year and expanding beyond protein structure prediction to incorporate other biological data. Ultimately, they aim to develop a general-purpose AI model for biology that can be applied to various use cases, including medicine.
Editor Notes: A Paradigm Shift in Biology Research
The development of AI language models for biology has the potential to revolutionize the field, offering new insights and possibilities for drug discovery and various applications in healthcare. EvolutionaryScale’s ambitious goals and commitment to scaling AI models show their dedication to driving innovation in biology. While the road ahead may be challenging, the potential benefits make it a worthwhile endeavor.
It’s exciting to see startups like EvolutionaryScale and established companies like DeepMind pushing the boundaries of what AI can achieve in the field of biology. As AI becomes more sophisticated and capable, we can expect to see significant advancements in drug development and other areas of healthcare.
GPT News Room is thrilled to follow the progress of EvolutionaryScale and other companies working on AI for biology. Stay tuned for more updates and breakthroughs in this exciting field.
Visit GPT News Room for the latest news and insights on AI and technology: [GPT News Room](https://gptnewsroom.com)
Source link
from GPT News Room https://ift.tt/6fIBaxl
No comments:
Post a Comment