2025-05-20 –, On Air
The GPT-NL initiative is a collaboration between TNO, SURF, and NFI. It aims to create a Dutch Large Language Model (LLM) and its community through the development of a state-of-the-art research facility and the pursuit of a sovereign and lawful language model. An ambitious vision with a lot at stake! In this session, we address our key challenges: securing licensed data, creating an energy-aware training, mitigating bias, and protecting private information.
The session is designed to be dynamic and interactive, combining presentations with open Q&A periods to foster an open discussion.
In today’s rapidly advancing AI landscape, Large Language Models (LLMs) are becoming an integral part of a growing range of applications. These models are deep neural networks capable of understanding and generating text based on extensive datasets. In recent years, LLMs have been applied in areas such as creative writing, software development, and autonomous agents. Their development relies on training with vast amounts of text data, enabling them to recognize complex patterns and contextual subtleties in language.
GPT-NL is a project funded by the Dutch Ministry of Economic Affairs, aimed at developing a sovereign Dutch-English language model. Built from the ground up, GPT-NL will be a transparent, legally compliant LLM aligned with European standards. This includes clear accountability in how data is sourced and curated, with a strong emphasis on addressing bias, ethical considerations, and the responsible acquisition of properly licensed data. Through this, GPT-NL aims to promote openness, transparency, and data privacy for its users. Grounded on scientific research, we are working to optimize the scope, training, and implementation of the model, ensuring that it is both efficient and effective.
This is your chance to learn more about the future of Dutch AI, explore the challenges and opportunities ahead, and actively contribute to the GPT-NL community.
Dr. Rer. Nat. Julio Alexandrino de Oliveira Filho is a Brazilian-Dutch geek scientist, absolutely passionate about science and technology. His main drive is the automated, model-driven design of complex and adaptive cyber-physical systems. In GPT-NL, Julio is a lead system architect, responsible for the implementation and execution of the curation and training pipeline.