AlphaFold Changed Science. After 5 Years, It’s Still Evolving (www.wired.com)

AlphaFold, Google's DeepMind AI, celebrated its fifth anniversary, marking a transformative period in the field of biology and chemistry. Launched in November 2020 and awarded the Nobel Prize in Chemistry last year, AlphaFold's breakthroughs in predicting protein structures have led to a comprehensive database containing over 200 million predicted protein shapes, utilized by 3.5 million researchers globally. With the recent introduction of AlphaFold 3, the AI has expanded its capabilities to include DNA, RNA, and drug interactions, tackling significant challenges like structural hallucinations while maintaining a focus on rigorous verification methods to ensure reliability. Looking ahead, DeepMind's vice president of research Pushmeet Kohli highlighted three key areas: enhancing AI's reasoning capabilities, democratizing access to these tools for scientists, and ambitiously pursuing the simulation of entire human cells. The emergence of an “AI co-scientist,” built on Gemini 2.0, is set to revolutionize scientific research by automatically generating and critiquing hypotheses, which enables human researchers to concentrate on formulating valuable questions. This partnership aims not only to accelerate hypothesis generation but also to deepen understanding of complex biological systems, paving the way for revolutionary advancements in personalized medicine and beyond.