Datasets:
License:
| license: cc-by-sa-4.0 | |
| tags: | |
| - opensource | |
| - opensourceai | |
| # Building up Open Source AI with the Hugging Face Community | |
| As the field of AI advances, it's becoming increasingly important to understand what "Open Source" means in the context of artificial intelligence. | |
| While Open Source software has long had established standards, AI models and datasets bring new challenges around transparency, accessibility and collaboration. | |
| To address this, the Open Source Initiative (OSI) has been developing an **Open Source AI Definition** (OSAID) which will outline the criteria that AI components—such as models, | |
| datasets and code—must meet to be considered Open Source. | |
| ## Co-designing the Open Source AI Definition v1.0 | |
| Understanding that AI impacts a global community with diverse perspectives, we embarked on the quest to define Open Source AI by adopting a co-design methodology—a | |
| participatory approach that shares knowledge and power among all stakeholders. This method allows us to integrate diverging viewpoints into a cohesive and consensual | |
| standard. | |
| The 1.0 version of the definition is set to be unveiled on October 28, 2024, at the [All Things Open](https://2024.allthingsopen.org/osi) conference. | |
| The definition lays out the main requirements for what we hope will become a robust standard for OSAI across the ecosystem. | |
| ## Open Source AI in practice at Hugging Face | |
| To support the Open Source AI Definition, we need to see how it will be operationalized in practice by the people building and using Open Source AI systems. | |
| What Open Source licenses and legal frameworks are favored for sharing model weights? How is training data re-used and documented? | |
| Given the role of the Hugging Face platform in showcasing those decisions, we invite the community to use this resource to share that information and discuss | |
| new related questions as they come up; so we can keep building up Open Source AI with a shared vocabulary and understanding. | |
| ## What is the purpose of the Open Source AI Definition (OSAID)? | |
| The OSAID outlines the core freedoms that should apply to AI systems to be considered truly Open Source. These include the right to: | |
| - **Use** the system for any purpose and without having to ask for permission. | |
| - **Study** how the system works and inspect its components. | |
| - **Modify** the system for any purpose, including to change its output. | |
| - **Share** the system for others to use with or without modifications, for any purpose. | |
| ## What are the benefits of Open Source AI? | |
| Open Source AI offers significant benefits, including enhanced transparency and safety by enabling system audits, mitigating bias, and ensuring accountability | |
| in data sources, which accelerates AI safety research. It also fosters competition and innovation by making more models available, challenging AI monoculture, | |
| and providing access to foundational technology for a wider range of stakeholders. Additionally, Open Source AI supports the development of diverse, localized | |
| applications that reflect different cultural and linguistic contexts, allowing models to be aligned with various value systems. | |
| # How can the Hugging Face community participate? | |
| As the largest platform for the AI community, Hugging Face plays a key role in promoting the Open Source principles in the context of AI. | |
| We would like to invite both individuals and organizations of the Hugging Face community to [review](https://opensource.org/deepdive/drafts/the-open-source-ai-definition-1-0-rc1), | |
| [endorse](https://opensource.org/osaid-endorse), and adopt the OSAID, and to [discuss OSAI artifacts](https://huggingface.co./datasets/opensourceorg/osaid/discussions/1) on this page. | |
| This is your chance to take part in a key moment for AI’s future. | |
| As we approach the release of the Open Source AI Definition v. 1.0, your support is fundamental. | |
| We encourage you to explore the definition, contribute models that adhere to the OSAID, and engage in discussions around Open Source AI. | |
| Together, we can ensure that Open Source AI continues to evolve in ways that are fair, transparent, and accessible to all. | |