Shared Concerns Need Common Responses
AI technologies have the potential to amplify existing global power imbalances, but they also offer a real chance to foster a more just and equitable world.
More than any technology before it, AI is at risk of being controlled by a small few. Only a handful of big tech companies possess the capital, data, and compute power required to develop cutting-edge AI systems. Only a few governments have the capacity to regulate these companies and even they struggle to do so effectively.
In addition, these same companies use their immense capital advantage to hire the best experts in the field, driving salaries to unsustainable heights and concentrating talent within their ranks.
Despite the rhetoric, it is implausible to expect those who hold this power to give it up voluntarily.
This is why it is crucial to build a strong coalition of all those who are sincerely committed to using AI in service of a more just and equitable world.
If we can pool our resources, capacities, and knowledge, we have a real chance to shift the balance of power towards more advanced, cooperative organizational structures that align with a sustainable and healthy future for humanity.
A Cosy Approach
With CosyAI, we invite everyone who shares these concerns to join forces and help build a shared framework for AI that benefits all by design, not just by promise.
We aim to achieve this through a cooperative approach, grounded in our common human capacity for trust, connection, and community values. (“Cosy” stands for cooperating sincerely.)
To make this real, we will establish a multi-stakeholder cooperative at the EU level (SCE), with open global membership for individuals and value-aligned organizations. This legal framework will allow us to define shared standards and ensure collective ownership.
To keep ownership and accountability as close to local communities as possible, we will also provide tools and legal support for the creation of local cooperatives. These will enable community-based ownership of computing power and data, while contributing to a scalable, shared infrastructure through technologies such as federated learning and decentralized server clusters.
Complementing Public Regulation
The challenge of climate change has shown how public institutions can struggle for decades to create and enforce the necessary global regulations. Regulating AI in a responsible way presents another urgent global challenge that demands balanced and well-conceived international responses.
It would be naive and irresponsible to rely solely on the hope that public institutions have the technical expertise and decision-making structures to act quickly and decisively enough.
At the same time, we are committed to working with public institutions in meaningful and constructive ways. We aim to support the development of balanced and effective regulatory frameworks for AI, in particular by contributing a perspective rooted in cooperating sincerely.
Cooperatively owned apps
CosyAI aims at offering its members AI-based apps that can be accessed through a common subscription model, thereby simplifying the use of safe AI for our user-members.
Cooperative LLMs
At the same time, we aim to train specialized cooperative AI models based on data provided by our members and co-owned through our cooperative structure.
Our goal is to develop unique, domain-specific models with strong data privacy guarantees. These models will enable members to deliver distinctive value to their customers while ensuring decentralized ownership and fair distribution of the value created.
Decentralized GPU clusters and data ownership
Two fundamental questions will strongly influence how the future of AI unfolds:
- Who owns and controls the hardware that AI models run and learn on?
- Who owns and controls the data needed to train increasingly capable AI models?
CosyAI aims to build the organizational infrastructure for cooperative ownership of decentralized GPU clusters and data.
We rely on technologies developed by our members and the broader open-source community, and we aim to use the most advanced tools available for decentralized GPU clusters, federated learning, data DAOs, high security standards, and data privacy. We are technology-agnostic and will adopt any tools that help advance our mission effectively and efficiently.
A Basic Income backed by AI
A third fundamental question about the future of AI is:
- Who benefits from the value generated by AI?
As another key step, we seek to create a basic income for people who join and contribute to building a more cooperative world.
The basic plan is simple: CosyAI will generate revenue by offering subscriptions to safe AI tools and services, as well as access to compute power for inference and model training.
A portion of the surplus generated by the CosyAI cooperative will be directed into a basic income fund, which will automatically distribute payments to accredited contributor-members.
Over time, both the sources of funding and the group of recipients will be expanded.
Building an AI Safe Space
Many of us share broader concerns that the rapid development of AI could outpace society’s ability to respond, exposing us to increasing levels of fake information, manipulation, job loss, and other risks we are not adequately prepared for.
This is why, in the long run, we aim to create an AI Safe Space for people who want to coexist with AI on their own terms without surrendering control to AI systems.
Our goal is to empower individuals and organizations to make a conscious choice about the extent to which they engage with advanced AI, and to ensure they can safely opt out if they choose to.
It’s time to make it happen – join or stay updated.