The First Step Toward a User-Driven AI Era Has Begun
On the 26th, Vana and Flower Labs announced their collaboration to develop "COLLECTIVE-1," the world’s first user-owned decentralized AI model. This model is trained on data voluntarily provided by millions of users, granting individuals full ownership and control over their data and its usage.
Unlike traditional AI models, which are primarily trained on publicly available web data, COLLECTIVE-1 leverages Vana’s personal data network and Flower Labs’ decentralized AI training technology. It incorporates unique personal data such as conversations, health records, location information, sleep patterns, and vehicle driving data. The two companies stated that this approach enables the creation of a more refined and personalized AI system.
Vana has established a "Data DAO" ecosystem, allowing users to contribute data and receive direct rewards. Currently, tens of millions of user-owned data points have been accumulated within the Vana network. Flower Labs, meanwhile, operates "Flower," the world’s largest open-source federated learning framework, and brings experience in training large language models with 100 million to 700 million parameters in a decentralized manner.
Through this partnership, the two entities aim to empower developers to build better AI models using real-world user data rather than relying on web crawling. They also plan to release technical documentation and examples as open-source resources to encourage widespread participation.
Anna Kazlauskas, founder of Vana, remarked, "While traditional AI models have allowed a handful of large corporations to monopolize value, COLLECTIVE-1 shifts this paradigm. A decentralized model co-created by millions possesses a data advantage that no single company can replicate."
Nic Lane, Chief Scientist at Flower Labs, added, "Decentralized AI training is the key to unlocking the next chapter of AI. COLLECTIVE-1 will set a new state-of-the-art (SOTA) standard, surpassing conventional centralized models."
