Serbia developing language model for Serbian language that opens up space for application of AI in education, healthcare, justice, culture, public administration and economy

The Director of the Office for IT and eGovernment, Mihailo Jovanovic, and the President of the Serbian Chamber of Commerce, Marko Cadez, signed a Cooperation Agreement with the aim of more efficiently implementing the program of spreading knowledge in the field of digital transformation "Serbia is Digitalizing", as well as developing a national LLM model, i.e. a large language model for the Serbian language, the Office for IT and eGovernment announced.

Jovanovic stated that this Agreement comes at a time when, according to the World Bank index measuring the maturity of public administration development, Serbia ranked 2nd in Europe and 6th in the world, while according to the Oxford Government AI Readiness Index, it was ranked among the 20% of the most prepared countries in the world. He emphasized that it is in such a context that this Agreement represents an important step towards the international promotion of domestic solutions in the field of eGovernment, as well as towards the further development and wider application of artificial intelligence.

- For the first time, Serbia is systematically and strategically approaching the development and training of a large AI language model for the Serbian language, which opens up space for the application of artificial intelligence in education, healthcare, justice, culture, public administration and economy. By developing a large AI language model for the Serbian language, we are preserving our digital and AI sovereignty, ensuring that the data remains in the country. This model will enable the Serbian language, script and identity to be equally represented in the global digital space, with full control over data, standards and ethical principles. Thanks to the existing supercomputer in the State Data Center in Kragujevac and the planned installation of two more supercomputers in 2026, Serbia is creating and improving its own AI infrastructure, while the development of the Serbian language model has great cultural and identity significance - said Jovanovic.

As he added, it is expected that many other institutions will soon join this project, primarily the National Library, the University Library, RTS, RTV, the Institute for the Development of Artificial Intelligence and many others.

Marko Cadez emphasized that the development of a national AI language model for the Serbian language will have enormous benefits for the economy, such as reducing the cost of developing solutions, increasing productivity, automating customer support and sales, as well as faster data-based business decisions.

- Every company, regardless of the level of digitalization, will be able to transform more easily, faster and simpler, with significantly lower entry barriers for the application of artificial intelligence. This applies to all sectors - from energy and agriculture, where AI enables smart resource management, through finance with advanced digital assistants and communication with clients in the Serbian language, to healthcare through the analysis of medical documentation, faster diagnostics and secure storage of citizens' health data - he stated.

In education, children and students, as he emphasized, will have access to AI tools in the Serbian language, with security filters, while in the legal system, the domestic language model will improve the work of lawyers, courts and notaries, with precise and standardized use of language.

- Small and medium-sized enterprises will experience the biggest change through automated sales, marketing assistants and digital business support, which will enable them to be more competitive in the domestic and international markets. Creating their own language model and introducing the 5G network are two key milestones that position Serbia as an exporter of innovations, a strong digital ecosystem and a regional innovation hub, with clear benefits for the economy, citizens and society as a whole - said Cadez.

He said that the plan is to internationalize IT solutions and market innovative tools developed by the Office for IT and eGovernment. Negotiations are underway with Tashkent to take over the eBeba program, as Uzbekistan is the youngest Central Asian republic, where a million babies are born annually.