Near Protocol (NEAR), a leading altcoin project, has embarked on a significant initiative in the world of artificial intelligence. Introduced at the opening of the Redacted conference held in Bangkok, this project aims to develop the largest open-source AI model in the world, boasting 1.4 trillion parameters, which is 3.5 times larger than Meta’s current open-source Llama model.
Decentralized AI Ensures Privacy
The development of the AI model will involve participants collaborating in a competitive environment through the Near AI Research hub. Starting today, participants eager to train a smaller model with 500 million parameters can join the project. The creation of this model requires a significant investment of 160 million dollars, as noted by Illia Polosukhin, one of the founders of Near Protocol. Polosukhin stated, “This cost will be covered through token sales, and token holders will be rewarded with revenues generated from the model’s usage.”
The project aims to offer a decentralized approach to the growing influence of artificial intelligence. This model, which lacks a central structure, will utilize encrypted Trusted Execution Environments to reward participants’ contributions and ensure privacy. Polosukhin emphasized that this new technology is crucial for addressing privacy concerns surrounding AI models.
Emphasizing Decentralization Against Central AI
Edward Snowden, who spoke at the conference, highlighted the risks of centralized AI. He warned, “If AI is controlled by a single company, the entire world could turn into a surveillance state.” Snowden advocated for a decentralized structure for AI, underscoring the importance of Web3 philosophy. Developing AI in a decentralized manner will not only protect users’ privacy but also allow for more transparent technological advancement.
Near Protocol’s move is considered a significant step towards a more free and transparent future in AI. As the impact of AI on society continues to grow, decentralized solutions hold the potential to minimize risks in this field.
