Near Protocol announced its plans for the largest open-source AI model at the Redacted conference in Bangkok.
Initiative Launch
Near Protocol has announced the development of a 1.4 trillion parameter AI model, surpassing Meta's Llama. The project is set to involve crowdsourced research facilitated by the new Near AI hub. Participants have the opportunity to engage in training a smaller-scale, 500 million parameter model starting November 10.
Project Development
The project will expand across seven different models, retaining the best researchers and contributors at each stage. The model intends to monetize through an encrypted Trusted Execution Environment to reward contributors. Co-founder Illia Polosukin mentioned funding of roughly $160 million will be sourced from the crypto market.
Challenges and Privacy Concerns
The project faces challenges such as acquiring numerous GPUs. Employing a decentralized network for computation will require currently nonexistent technology. However, Deep Mind's research shows feasibility. Polosukin is open to partnering with the Artificial Superintelligence Alliance. Guest speaker Edward Snowden stressed the importance of decentralized AI to avoid surveillance concerns associated with centralized models.
Focusing on decentralized AI can significantly alter the current approach to privacy and security.
