First Serbian AI language model to be presented by 2027

Director of the Office for IT and eGovernment Mihailo Jovanovic said that Serbia, at a time when the whole world was competing for leadership in the era of artificial intelligence, was setting ambitious goals, which included storing its data in the country, as well as developing its own systems and software based on artificial intelligence.

He told RTS that in the previous decade, Serbia had laid a solid foundation for the development of digital infrastructure by building the State Data Center in Kragujevac and purchasing the first supercomputer, which was also located in Kragujevac.

– In accordance with the Serbia 2030 strategy, a significant expansion of the existing capacities of state data centers in Kragujevac and Belgrade is planned, as well as the construction of two new state data centers in Nis and Novi Sad. The goal is for the total power of all state data centers in Serbia to reach more than 200 megawatts by 2030, which represents a significant increase compared to the existing capacities – said Jovanovic.

He stated that the first supercomputer in Serbia was already largely at capacity, which is why the installation of the second supercomputer was underway, which he expected to be put into operation in April.

– It will be as much as seven times faster than the existing one. At the same time, we also expect the ratification of the agreement in the National Assembly for the procurement of the third supercomputer, which should arrive by the end of the year. With three supercomputers, Serbia will have a very significant, perhaps one of the largest capacities in this part of Europe – said Jovanovic.

He added that Serbia, thanks to supercomputers and domestic experts, had the capacity to develop its own language model based on artificial intelligence.

– In this process, we will also use the support of the French company Mistral, one of the fastest growing technological startups in Europe. We will cooperate with their engineers to accelerate the development and implementation of these technologies – said Jovanovic.

According to him, it is important that citizens understand the importance of developing a domestic language model.

– Tools like ChatGPT and other global systems use the amount of Serbian language and data that they can find on the Internet. It is important for us to include our cultural heritage, literature, scientific works and everything that makes up the richness of the Serbian language in the development of our own model, so that artificial intelligence can understand our language, context and culture. The plan is to present how the Serbian language model works by 2027, when one of the largest international events awaits us - Expo 2027 – said Jovanovic.

Kancelarija za informacione tehnologije i elektronsku upravu Beograd

Mistral

Mihailo Jovanović

Serbian AI language model

artificial intelligence