Pak will help to set this topic, but both of them and Zoo wanted to see that the virtual lab can come up with it. As a first project, he decided to focus on the new Coid -19 stress treatment. In keeping with this goal, Zoo trained five AI scientists (who have been trained to act like an immunologist, a computational biology, and a principal investigator), in which they have been trained with different goals and programs.
It took a few months to build these models, but Pak says that once the setup was completed, they were very quick to design candidates for treatment: “I think it was a day or a half day, something like that.”
Zhu says agents decided to study anti -quad nanobes, a cousin of antibodies that is very small in size and less common in wild. Zoo was surprised, though, for that reason. They claim that the models landed on Nanobes that after contacting these small molecules would be in accordance with the limited computational resources that were given the models. “This was actually a good decision, because the agents were able to design these nanoboes effectively.”
According to the study, the models developed were actually new progress in science, and most of the original Quaid -19 were able to bind different forms, according to the study. But both Pak and Zoo admit that the basic contribution of his article is really a virtual lab as a device. Yi Shi, a pharmacologist at the University of Pennsylvania, who was not involved in the work, but changed some of the basic nanobides modified to the Virtual Lab. They say they like the Virtual Lab demonstration and that “the biggest novelty is automation.”
Nature Accepted this article and tracked it fast for a preview of the publication-you knew that taking advantage of AI agents for science was a hot area, and he wanted to join the first people to test it.
AI scientists hosted a conference
When he was offering his dissertation, Zoo was afraid to see that he could not credit the AI ​​properly for his role in research. Most conferences and magazines do not allow AI to be listed on the papers as co -ordinators, and many researchers forbid the use of AI to write papers or reviews. NatureFor example, References Uncertainty about mistakes in the reasons for accounting for accountability, copyright and its practice. “I think it’s limited,” says Zoo. “This type of policies are encouraging researchers to basically hide or minimize the use of AI.”
Wanted to turn the script by forming zoo Agents 4 this science conferenceIn which all the subordinates need to become AI. Other boats will then try to evaluate the work and determine its scientific qualities. But people will not be completely released from the loop: a team of Nobel Laureate, including human experts in economics will review top papers.