In April 2025, this commentary from Derek Lowe at Science discusses recent optimistic claims by Google DeepMind's CEO about using AI to revolutionize and potentially end disease within a decade by rapidly advancing drug development through protein structure prediction. Lowe expresses strong skepticism towards this vision, arguing that current AI, primarily machine learning, only identifies patterns in existing data and does not generate new fundamental knowledge needed to tackle biological systems and disease complexities. He contends that predicting protein structures, while a significant achievement, addresses only a small part of the multifaceted challenges in understanding and curing diseases, highlighting the vast unknown aspects of biology. Lowe suggests that while AI might aid research, substantial discoveries and insights from human scientists are still essential before such ambitious goals can be realized. Produced by Dr. Jake Chen.