The expert assessed the possibility of replacing scientists with artificial intelligence

Shanghai Jiao Tong University has developed an autonomous biological laboratory called autoDNA under the control of AI agents. As reported in the journal Nature, the system is capable of independently planning and conducting experiments based on a text query, controlling more than 20 devices, performing thousands of operations, synthesizing and sequencing DNA.

However, it is premature to talk about the complete replacement of scientists with artificial intelligence. This was stated by Alexander Panov, head of the Center for Robotics and Unmanned Technologies at the MIPT Institute of AI, in an interview with Nauka Mail on Monday, November 9. According to him, automation of biological and chemical laboratories has been developing for a long time, and the greatest success has been achieved in DNA sequencing due to the uniformity of operations.

However, human involvement is still necessary to replace reagents, maintain equipment, and perform non-standard procedures, including weighing substances and loading samples into furnaces or centrifuges. In addition, specialists should check the results of automated systems, as they may make mistakes.

Panov noted that in DNA research, additional automation usually does not increase the number of errors due to strict procedures and verification of results. At the same time, AI and robots help biologists get rid of routine tasks rather than replace them. According to the expert, automation may affect Computer Science specialists earlier, since their work does not require interaction with the physical world.

The expert stressed that artificial intelligence is not yet able to independently engage in science. Although individual language models already perform mathematical calculations and prove theorems, they still act primarily as assistants in biology and chemistry.

Dose of learning: AI will be able to more accurately detect Alzheimer's disease in the early stages

Russian technology has helped scientists from Harvard to compensate for the lack of genetic data of patients

In October, Pavel Balabanov, chief researcher at the Institute of Higher Nervous Activity and Neurophysiology of the Russian Academy of Sciences, said that the use of AI in research can lead scientists to a dead end due to the high proportion of algorithm errors. 360.ru . One false result can distort the direction of all subsequent experiments, writes NSN.

At the same time, cardiologist, gerontologist, and internist Yuri Serebryansky expressed the opinion that with the expansion of the use of AI in surgery, control over operations will remain with humans. He clarified that AI is already being used in the treatment of cardiovascular and oncological diseases, in transplantation and vascular surgery, RT notes.

Also in October, it became known that AI Institute scientists have developed LongTails, a software platform that automates a significant part of mathematical research. The system formulates mathematical optimization problems, looks for ways to solve them, and formally proves the results.

Prior to this, it was reported that the Moscow Institute of Physics and Technology (MIPT) has opened a postgraduate course, which teaches the creation of advanced AI assistants and building a personal AI environment. As part of the course, graduate students are taught to integrate AI tools at all stages of scientific work, verify the results of neural networks, and maintain authorial control over research, the Moscow City News Agency writes.