Joachim Frank says AI can speed science but not replace experiments

Nobel laureate Joachim Frank said artificial intelligence is becoming a powerful tool in scientific research, but experimental methods remain essential for testing hypotheses and producing new evidence. The interview highlights how AI and lab work are increasingly shaping fields from structural biology to drug discovery.

Why it matters: - AI is changing how scientists process data, generate hypotheses and search for patterns across massive datasets. - Frank’s comments underscore a key limit: computational tools can assist discovery, but experiments still validate what science can claim. - The balance matters in fields such as structural biology, drug discovery and materials research, where speed and accuracy both shape outcomes.

What happened: - Southern Finance and 21st Century Business Herald interviewed Joachim Frank, the 2017 Nobel Prize laureate in Chemistry, about AI and the future of experimental science. - Frank said AI can accelerate parts of the research process by analyzing scientific data and helping researchers develop hypotheses. - Frank also said experimental work remains essential for testing hypotheses and studying phenomena that are not yet understood. - The interview was part of Southern Finance’s “High-Level Interviews: Conversations with Global Scientific Giants” series.

The details: - AI systems can process large datasets, identify patterns and support tasks that once required substantial time and computing power. - Frank said AI relies on existing knowledge and experimental data. - Scientific research also depends on observations and experiments to test new hypotheses and generate new evidence. - Frank pointed to his own work on single-particle cryo-electron microscopy as an example of computational methods and experiments advancing together. - His approach helped determine the three-dimensional structures of biological molecules from large numbers of two-dimensional electron microscopy images. - The method addressed a major problem in structural biology for molecules that were difficult or impossible to crystallize. - By combining electron microscopy with computational image processing, researchers could reconstruct molecular structures from noisy individual images. - Frank won the 2017 Nobel Prize in Chemistry with Jacques Dubochet and Richard Henderson for developing cryo-electron microscopy for high-resolution structure determination of biomolecules in solution. - Frank’s research has also been connected to the development of cryo-EM in China. - Chinese researchers including Gao Ning, Lei Jianlin and Gao Haitao worked in Frank’s laboratory and later contributed to cryo-EM research and training in China.

Between the lines: - Frank’s view reflects a broader scientific shift: AI is becoming a force multiplier, not a substitute for the lab. - The message also suggests that the most durable scientific advances may come from combining computation with direct observation rather than treating them as competing approaches.

What's next: - AI adoption is likely to keep expanding in research areas where large datasets and complex modeling can improve efficiency. - Experimental methods will remain central as researchers use them to confirm results, refine models and explore new questions. - The interview adds to ongoing debate over how much of the scientific workflow can be automated without weakening the evidence base.