Researchers are exploring advancements in protein structure determination, which previously required substantial financial investment, up to $100,000 per structure. Jumper highlights the aim to enhance structure prediction’s role in the research process by leveraging groundbreaking technology, suggesting an approach akin to “making everything into nails” to maximize efficiency and impact.
Looking ahead, Jumper plans to integrate the precision of AlphaFold, a system for protein structure prediction, with the capabilities of large language models (LLMs). He notes that current technologies can analyze scientific literature and perform some level of scientific reasoning. The goal is to create advanced systems that merge these two methodologies for better protein structure predictions.
Another initiative, known as AlphaEvolve, developed by a different group at Google DeepMind, employs an LLM to generate potential solutions and utilizes a second model for verification, eliminating less viable options. This system has successfully led to discoveries in mathematics and computer science; it raises questions about whether Jumper envisions a similar synthesis of LLMs with his work.
When asked about future methods, Jumper refrained from providing specifics but expressed anticipation regarding the expanding influence of LLMs on scientific research. He acknowledges the speculative nature of this topic yet identifies it as an intriguing area to explore.
At 39, Jumper holds the distinction of being the youngest chemistry laureate in 75 years and reflects on his career’s midpoint. He emphasizes the importance of pursuing smaller, innovative ideas rather than aiming solely for monumental achievements, stating that future announcements need not replicate the magnitude of his Nobel Prize-winning work.
Source: https://www.technologyreview.com/2025/11/24/1128322/whats-next-for-alphafold-a-conversation-with-a-google-deepmind-nobel-laureate/