Evo 2 - A Generative AI Model That Advances Genome Research.

In recent years, rapid advancements in technology, particularly in artificial intelligence, have become common. AI is now being integrated across every sector, from agriculture to intelligence services, from academia to government, institutions and companies around the world are developing AI systems tailored to their specific tasks and needs.

A few months ago, a team of scientists from Stanford University, NVIDIA, and the University of California (Berkeley and San Francisco) announced the development of a generative AI model called Evo 2. This AI was trained on the genetic datasets of a wide range of living species. Evo 2 can generate genetic sequences and predict the structure and function of the proteins that those sequences would produce.

The researchers claim that this technology has major implications for understanding diseases and addressing global challenges such as environmental crises. DNA itself is a kind of biological code. It is composed of four nucleotide bases—adenine (A), thymine (T), guanine (G), and cytosine (C), which function like letters in a language. These bases pair and sequence in specific ways to form genes, just as letters combine to form words. Then, those genes are translated into proteins, the molecules that carry out most biological functions.

Evo 2 is a generative AI, and they are similar in principle to language models such as ChatGPT. Whereas ChatGPT is trained on large datasets of human language and can generate text, Evo 2 is trained on genetic sequences from many species and can generate new genetic codes. It can also predict how mutations might alter proteins and how these changes may contribute to diseases such as cancer.

However, the researchers warned that this capability could be misused such as to create new, potentially harmful genetic variants, giving emphasize the need for ethical and safety guidelines as such technologies evolve.