A team of researchers at Estonia’s University of Tartu (TÜ), Tallinn University of Technology (TalTech), and the Institute of the Estonian Language (EKI) will teach the Estonian language and culture to large language models (LLMs), ERR reports. The project led by TÜ’s Institute of Computer Science will train open-source LLMs to speak Estonian more fluently and better understand the Estonian culture. Kairit Sirts, associate professor in natural language processing at TÜ, noted that Estonian in AI applications now often sounds artificial. The project will train the LLMs on the LUMI supercomputer in Finland.