The expert explained the inability of AI to make discoveries in chemistry

Valery Fokin, professor at the University of Southern California, director of the Center for Biomolecular Research and winner of the Challenge Award, explained why artificial intelligence (AI) is not yet capable of making discoveries in chemistry.

In an interview with the Science Mail on Tuesday, August 4, the professor said that AI is already helping chemists design new molecules and search for promising compounds, but it is not yet able to make scientific discoveries on its own.

According to the scientist, the main problem is not the algorithms, but the lack of high-quality data for their training. Now AI can predict the structure of molecules, but it cannot yet reliably determine what properties they will have and whether they can become, for example, medicines or new materials.

Fokin also noted that AI will not replace researchers in the foreseeable future. According to him, today scientists are paying more and more attention to the quality of experimental data, rather than the development of algorithms themselves. Of particular importance, the expert stressed, are the negative research results, which are still insufficient in chemistry and drug development. He believes that the accumulation of reliable experimental data will be the main condition for the further development of AI in this area.

Between lines and neurons: Can AI replace the human brain

The research opens up new opportunities for the development of AI, neurointerfaces, and medicine.

On the eve of August 3, Alexander Ginzburg, Academician of the Russian Academy of Sciences, scientific director of the N. F. Gamalei Research Center for Epidemiology and Microbiology, said that three new cancer vaccines would be created in Russia in the future. Prior to that, Russian scientists had developed a personalized antitumor mRNA vaccine, Neoonkovac, which was created using artificial intelligence technologies.

On August 1, Alexander Kugaevskikh, associate professor at the Faculty of Software Engineering and Computer Technology, head of the Master's degree program "Design and Development of AI Systems" at ITMO University, told Izvestia that modern AI systems consume significantly more energy than the human brain, because they use a fundamentally different approach to information processing. The expert also stressed that the principle of operation of biological neurons differs significantly from artificial intelligence algorithms.

In June, Veronika Skvortsova, head of the Federal Medical and Biological Agency of Russia, told RT that with the help of artificial intelligence, scientists are finding mutations in the human genome to create a vaccine against tumors. The results of sequencing the RNA of a human tumor and its blood are used as the material for analysis, notes 360.ru .