Datasets:
From data collective to public AI cooperative
Initially I set up this discussion in a /collective dataset, with the purpose of supporting the efforts of Public AI on a local level. This is (among other things) an experiment in collecting user data in a consistent way from the inference utility launched with Apertus. Discussions with our pioneer community have made the purpose of the project more clear, and we created a basic website & renamed this Hugging Face org today. We are currently mostly active on Slack, GitHub and BigBlueButton, but I am encouraging folks to move some threads to the open forum here.
In the last meetings we have focused on the proposal of creating a cooperative that would pool together resources and provide governance around Inference Utilities. I had a brief chat with Apertus to help explain this:
- Collective: Refers to groups of people who come together to achieve a common objective. It may be a self-organizing group, or it may be formally structured as a legal entity, such as a union or a social enterprise. Collections can act as one single entity, with a common identity, goals, and methods.
- Cooperative: While a collective is a group of people united by a common goal, a cooperative is a type of collective that works together to achieve a mutual goal. Cooperatives are typically structured as legal business entities, and membership is open to everyone who shares a common interest (sometimes with requirements for voting or holding office). Benefits are usually distributed equitably among members, based on their usage or contributions.
To support and sustain the mission of Apertus in Switzerland effectively, the Swiss Public Inference Utility (SPIU) is being launched as a cooperative model. This cooperative model offers several advantages:
- Local Focus: Addressing the social impact goals of public AI in Switzerland, which can lead to tailored public services.
- Community Engagement: Spreading the financial, political, and technical responsibilities among members can foster a committed and more resilient community.
- Decision-Making: Swift decisions can be made, enabling the cooperative to adapt and evolve to address changing demands and challenges.
Moreover, a cooperative structure aligns well with Swiss values of inclusivity, responsibility, and community engagement. It provides an opportunity for individuals and businesses to come together, share resources, and work towards a mutual goal—creating a sustainable and transparent ecosystem for public AI in Switzerland.
The cooperative model empowers all participants to be a part of the future of public AI in Switzerland, offering a unified, reliable, and transparent service where users understand data handling and governance transparently. Every step of the way, the SPIU cooperative is committed to ensuring that the benefits of public AI are accessible, equitable, and responsible. For this reason, we continue to collaborate with the Swiss AI Initiative, universities, governments, and research centers to develop best practices and ensure the ethical, secure, and fair use of public AI models in Switzerland.
We look forward to welcoming you to the SPIU cooperative and thank you for your continued interest in shaping the future of public AI in Switzerland. Let's build toward a more inclusive, responsible, and community-driven use of AI, together.
Any questions? Thoughts in particular on how to structure & make the most of our presence on Hugging Face?