Shedding light on…n°7 – Open source and AI: rethinking synergies?

Open source, transparency, openness: these concepts are becoming increasingly commonplace in AI speeches. The unprecedented increase in the capabilities of models, especially generative ones, raises many questions about their training and distribution framework. This new Éclairage sur... explores licensing formats and how they match the needs of stakeholders.

Generative Artificial Intelligence has recently experienced a dazzling breakthrough, bringing the automatic production of synthetic content (text, sound, images, video) within everyone's reach. Deep learning models are the cornerstone of this booming industry. Created from a vast amount of data of varying status, they represent software building blocks with a wide range of applications, whose format for dissemination or eventual reuse raises many questions.

Today, the main players in this new ecosystem rely directly or indirectly on resources available under open source licenses. In order to understand how these innovations have come to be based on such a foundation, and to assess whether this framework remains suitable for their development, PEReN explores licensing formats in this issue of "Éclairage sur...", and how they can foster a sustainable and balanced innovation dynamic.

These files may not be suitable for users of assistive technology, you can request an accessible format