Israeli Scientists Reveal Evolution May Follow More Predictable Paths Than Previously Thought

Directed Drift Shapes Evolutionary Paths

According to НВ — Техно: A team of international researchers from Israel's Technion – Israel Institute of Technology unveiled new findings published in PNAS on September 14, 2026, that challenge the long-held belief that evolution is largely random. Utilizing a novel mathematical framework, the study explores how populations survive under natural selection and mutations across flat evolutionary landscapes.

The researchers discovered that populations tend to drift purposefully toward plateau regions of the landscape, even when these do not offer immediate survival advantages. This directional drift arises from the interplay between population variability and the geometric features of the evolutionary landscape. Flat regions, as the study indicates, provide a protective buffer that shields organisms’ traits from harmful mutations and environmental disturbances.

Implications for Evolutionary Theory and Artificial Intelligence

These insights prompt a reconsideration of traditional views that evolutionary changes are purely random or neutral. The team also highlights the potential applications of their mathematical models in artificial intelligence, particularly deep learning systems. They note that neural networks often favor flat areas within the solution space during training, mirroring evolutionary dynamics.

Looking ahead, the researchers aim to extend their models to analyze more complex biological ecosystems in the wild, opening new avenues for understanding evolutionary mechanisms. This research not only advances evolutionary biology but also holds promise for enhancing AI technologies by applying evolutionary principles to improve neural network performance.

The findings could transform how evolution is perceived, suggesting it operates with greater structure and predictability. This shift in understanding carries significant implications not only for biology but also for developing more efficient algorithms in artificial intelligence that mimic evolutionary learning processes.

These recent findings align with previous research that suggests evolutionary processes are influenced by genetic pathways established by common ancestors. By exploring how these genetic 'highways' guide animal evolution, scientists are uncovering layers of complexity in evolutionary dynamics. For a deeper understanding of these mechanisms, read more about how shared ancestry impacts species development in our detailed article on genetic influences on evolution.