It's 3:00 AM. In the emergency room of a major hospital, patients arrive one after another with similar symptoms: high fever, sudden exhaustion, and rapidly worsening respiratory distress. At first, doctors think they are dealing with a new wave of influenza, but the usual tests come back negative. Hours pass, then days, and hospitals in other cities begin reporting the same cases.
Hot Topics
The samples are sent to central laboratories, and there the surprise unfolds. The genetic material isolated by the researchers belongs to a virus never before recorded in any known database. It bears no clear match to viruses known to scientists, and in the initial hours, no one can determine its origin, the animal it may have jumped from, or even whether it arose naturally. As the infection spreads, another, more dangerous race begins than the doctors' race against the disease. Intelligence agencies search for the source of the outbreak, governments close borders, and social media is flooded with accusations and counter-accusations.
Then the researchers discover something even more alarming: the virus's designers didn't need years of experimentation to write its genetic code; they used an artificial intelligence model that learned the "language of life," in much the same way that chatbots learn human language, and then asked it to suggest a genome that hadn't existed before.
At first glance, it seems like something out of a science fiction film, but its foundations are already a reality. In a controversial scientific experiment, artificial intelligence designed 16 new viruses, previously unknown in nature, and these microorganisms were able to reproduce in the laboratory at a faster rate than their natural counterparts. Such a scientific achievement would have been the stuff of science fiction just a few years ago. However, a new study published in the journal Science on August 6th revealed that its authors were able to train AI models not only to detect viruses already existing in our world, as we are accustomed to, but also to create entirely new designs.
Researchers designed the complete viral genome using two generative artificial intelligence models, named "Evo 1" and "Evo 2," which were trained during the deep learning process to analyze genetic codes instead of words. Just as AI models in chatbots are trained on written texts to recognize and master language, the scientists used the same trick and provided "Evo 1" and "Evo 2" with what they called "the language of life."
This language is actually a huge database that includes diverse genetic codes, millions of genomes of different life forms. In this way, artificial intelligence models learned the basic rules that determine how DNA changes and evolves in living organisms. After they became familiar with the rules and how genetic mutations occur, they were asked to simulate this process by creating a complete genome. Thus, scientists designed 16 types of phages, which are viruses that can eliminate specific types of bacteria.
Tags:
technology
Fact
ReplyDelete