If a benchtop DNA synthesizer that can synthesize DNA with 5,000 bases is commercially available in 2030, what will be the cost of the cheapest model (in 2024 US dollars)?
MetaculusFEB 14
Researchers at the University of Washington have successfully created artificial intelligence (AI)-designed enzymes capable of performing complex multi-step reactions, specifically creating serine hydrolases that can break down ester bonds through a six-step chemical process.
The team combined multiple AI approaches, including RFdiffusion for initial protein generation and a new tool called PLACER (Protein-Ligand Atomistic Conformational Ensemble Resolver), which predicts how atoms should arrange themselves based on physical and chemical principles.
The engineered enzymes demonstrated catalytic activity 60K times better than previous designs. 18% of the final designs showed successful catalytic activity while 1.6% completed a full catalytic cycle without getting stuck, known as multiple turnover catalysis.
This study is transformative as it harnesses machine learning to design what only mother nature was capable of in the past. These findings could revolutionize the world and address environmental and health challenges, as AI-generated enzymes could be engineered to degrade plastics, capture carbon dioxide, or combat cancer with unprecedented speed — offering solutions to some of our most pressing global issues.
While the idea of curing cancer and saving the planet is inspiring, AI-enhanced biotechnology should be pursued with caution. Such tools could also be used to hack DNA, leading to serious privacy breaches where personal genetic data might be exploited. Biotech companies might also cross ethical lines by manipulating human DNA and causing unintended genetic changes or ecological disasters. The opportunities must be balanced by the risks.