A group of Polish research centres will launch a large language model (LLM) powering the famous ChatGPT artificial intelligence (AI). The Polish Large Language Universal Mode (or PLLuM) project aims to become a Polish-speaking intelligent assistant providing public services. 

PLLuM is being developed by the Research and Academic Computer Network (NASK) which is part of a consortium of several institutions under the supervision of the Wrocław University of Science and Technology. Its creators expect it to soon become known worldwide as the “Polish ChatGPT”. 

According to NASK, existing AI systems, such as Google’s Bard and ChatGPT have two main problems: they are not cost-free, and their algorithms cannot be examined or modified by their users. In addition, they were programmed using little Polish-language content.

Jan Kocoń, an AI scholar at Wrocław University, believes that many answers provided by ChatGPT concerning Polish culture, customs, and facts are based on non-Polish-language sources, which leads to “grammatical and stylistic errors.” 

NASK announced its willingness to “increase the visibility of [the Polish] language and culture […] marginalised in the currently available model” via “a significantly greater share of texts originally written in Polish and containing information about Poland (Polish science, art, history, law, economy, and others)”. 

The first version of the PLLuM is to be made available for open testing by Spring 2024.

Image: Unsplash

Author: Sébastien Meuwissen