In the last decade, the field of artificial intelligence (AI) has experienced unprecedented growth and advancement, much of which can be attributed to the open research culture and the free sharing of both discoveries and code. However, there has been a notable shift in how some of the most influential companies in AI, like OpenAI, have approached the issue of open versus closed AI. This shift highlights an interesting evolution in the balance between promoting open innovation and protecting commercial or sensitive developments.

Originally, OpenAI was founded with the promise of keeping its advancements in AI open for everyone's benefit. Over time, especially with the development of highly advanced large language models (LLMs) like GPT-3, OpenAI has adopted a more closed approach.

Despite the trend towards closed code in some sectors, the open-source spirit remains very much alive, evidenced by the release of models like Grok by xAI and LLaMA by Meta AI. These models represent an ongoing commitment to accessibility and collaboration, demonstrating that it is possible to advance cutting-edge AI research while openly sharing achievements.

The Hugging Face platform has been instrumental in this context, serving as a central repository for open-source models, including Grok and LLaMA, and facilitating collaboration and knowledge sharing in the global AI community. Hugging Face has not only promoted innovation through open access to pre-trained models and datasets but has also fostered an ecosystem where researchers, developers, and enthusiasts can collectively contribute and improve the state of the art in AI.

As some companies choose more restrictive paths to protect their innovations, open-source models ensure that the barrier to experimenting, innovating, and applying AI remains low, allowing for a broader distribution of its benefits.

In a conversation with Yann Lecun at the World Economic Forum, the importance of maintaining AI research and development in an open-source environment is highlighted. It is underscored that the main reason we have witnessed such rapid progress in the AI field over the last decade is the practice of open research. As soon as a result is obtained, it is published on preprints like arXiv and the source code is shared, allowing everyone to reuse it.

This balance between open and closed code in the AI ecosystem illustrates a complex dynamic between innovation, collaboration, security, and commercialization. The ongoing evolution in this space will be determinant in shaping the future of artificial intelligence, ensuring that its advances remain accessible and that its impact is fairly distributed across society.