The full interview is available in Czech as an audio recording on the Czech Radio Plus website.
Source: Czech Radio Plus, interview by Tatiana Čabáková. Translated from Czech.
plus.rozhlas.cz (CZ original)"I can only speak for artificial intelligence – but in many of its subfields, the Czech Republic ranks among the top countries in Europe," says internationally recognized scientist Josef Šivic. After 15 years of working abroad at prestigious universities and technology giants, he returned home to Prague, where he now works at the Czech Institute of Informatics, Robotics and Cybernetics (CIIRC CTU).
"If we count publications in prestigious journals produced by CTU alone, we are among the top five institutions in Europe. In robotics and machine learning, we are somewhere among the first seven or eight. So we are doing very well," the scientist assesses.
Šivic and his teams are involved in several research projects: from machine learning for robots in factories to applications in biomedicine. Using artificial intelligence, they are also working on improving a drug designed to dissolve blood clots.
"For example, in the case of Alzheimer's disease, we still do not fully understand everything. We understand some of the mechanisms behind the development of the disease to a certain extent, but our goal is to understand its molecular basis. Artificial intelligence is developing tools – also for other diseases – that will allow us to accelerate our scientific work, especially by making it faster," he describes a project carried out in cooperation with the team of Jiří Damborský from Masaryk University in Brno.
From weeks to seconds
Šivic also gives other examples of how AI is changing scientific work.
"If we want to study the interaction between two proteins, it can be calculated, but it takes a very long time. Hours, days, sometimes weeks. Machine learning makes it possible to accelerate the process using examples of similar interactions and learn to predict the result much faster, which helps, for example, in the development of new drugs."
"An analysis that could easily take months can be completed within minutes. A simulation that used to take hours or days can then be finished in a tenth of a second. All with the help of machine learning."
People are already using AI today, especially language models. According to Šivic, new tools are emerging that can help researchers work with scientific literature.
"There are already tools that can prepare research reviews based on a given topic, meaning they search through scientific articles on the selected subject. This is also a major revolution in research. It is still at the beginning, but it has great potential and, once again, mainly offers the possibility of speeding things up," he explains.
"Artificial intelligence is sometimes compared to electricity, which enabled us to amplify – to increase – our strength and our ability to perform physical work. What is interesting about AI, however, is that it also increases our ability to process information. Therefore, we will soon see accelerated progress in other scientific fields as well, which may lead to breakthrough discoveries. I mean fields such as physics, biology, and chemistry," he says.
According to the scientist, it is clear that currently used artificial intelligence models require large amounts of energy to operate. However, research is already addressing this challenge, and significant changes are expected in this area as well in the near future.
Photo by Kateřina Cibulka, Czech Radio Plus / Český rozhlas Plus