Multimodal Large Language Models have become a dominant paradigm in artificial intelligence, driven primarily by large technology companies with access to extensive proprietary datasets and computational resources. While these models enable impressive capabilities, they also raise fundamental questions for academic research in Europe. How can scientific innovation and openness be preserved when foundational models are developed largely in private settings? How can European values such as transparency, fairness, and sustainability be upheld under strong computational constraints? And how should academic research position itself when pretraining at scale is no longer broadly accessible?
This Speinshart Seminar brings together academic researchers working on multimodal foundation models with industrial and applied research experts for a week of focused reflection and dialogue. Rather than presenting finished results, the seminar is designed as a space for early thinking on methodological, conceptual, and strategic directions for European AI research in the era of proprietary multimodal models.
Please note: This is just an information regarding events taking place at SSC; public attendance is therefore not possible due to the character of the retreat.
