GPT-NL: Towards a Public Interest Large Language Model
conference paper
Artificial intelligence (AI) is evolving into a transformative force across various sectors. Large language models (LLMs) are at the forefront of this transformation. To date, LLMs are predominantly controlled by multinational, private companies that are misaligned with public interests. This monopolization threatens sovereignty. OpenAI’s ChatGPT is paradigmatic for this observation. In response, the Dutch government has invested in projects that foster sovereign innovation in numerous sectors. This paper introduces one of these projects – GPT-NL – aimed at developing a Dutch LLM and fostering a supportive ecosystem involving academic institutions, companies, and government. Though formally GPT-NL’s development is directly aligned with the European Union’s Guidelines on Trustworthy AI, in the scope of this academic work we focus on its alignment to requirements for public interest AI (Züger & Asghari, 2023). As such, GPT-NL serves as a case on public interest AI for future research that provides knowledge to practitioners.
TNO Identifier
1002098
Source title
2nd Workshop on ‘Public Interest AI’ co-located with the 47th German Conference on AI (KI 2024) , September 23, Julius-Maximilians Universität Würzburg