AI and machine learning have made significant strides in biology, from designing enzymes to combat plastic degradation to developing proteins that neutralize snake venom. However, experts caution against relying solely on AI to understand complex biological processes.
A recent study, conducted by researchers at Heidelberg University, investigated the use of AI software packages to predict gene activity in cells under different conditions. The results showed that these systems failed to outperform a simplified method of prediction, highlighting the limitations of AI in biology.
The complexity of biology cannot be reduced to a single aspect, and developing AI systems that work for one area does not guarantee success in others. This study serves as a reminder of the need for caution when relying on AI in biological research.
While other studies have reached similar conclusions, this particular approach provides an accessible explanation of the researchers’ methods, allowing readers to better understand the findings and their implications for the field.
Source: https://arstechnica.com/science/2025/08/some-ai-tools-dont-understand-biology-yet