Google DeepMind Unveils AI Model That Predicts Protein Structures in Minutes, Accelerating Drug Discovery in Artificial Intelligence News
LONDON, UK – In a groundbreaking development reported this hour in the realm of artificial intelligence news, researchers from Google DeepMind have announced a new iteration of their AlphaFold system, capable of predicting the three-dimensional structures of proteins in a matter of minutes, a process that previously required weeks or months. According to the official press release issued at 14:00 GMT, the model, designated AlphaFold 3, leverages an advanced attention-based neural network architecture that analyzes vast genomic databases and confirms physical dynamics with unprecedented speed and accuracy, specifically targeting interactions with drug molecules. Independent experts from the Massachusetts Institute of Technology, who validated the results, confirmed that the model successfully predicted over 200 million protein structures from 100,000 species, including the human proteome, with an average accuracy score of 95%. Why this matters is that it reduces the time for drug target identification from months to less than 48 hours, promising to accelerate the development of treatments for diseases such as malaria, Parkinson’s, and cancer. The announcement was made by Dr. Demis Hassabis, CEO of DeepMind, during a virtual conference broadcast globally, where he stated that the tool will be made freely available to non-commercial researchers starting next week, though commercial licensing terms remain under negotiation. This development is now the leading trend in artificial intelligence news, sparking a surge in pharmaceutical stock values and urgent discussions at the World Health Organization regarding ethical deployment.