Near Protocol recently announced an ambitious attempt to build the world’s largest open-source AI model with an astonishing 1.4 trillion parameters. Announced at the Redacted conference in Bangkok, Thailand, this model will surpass Meta’s Llama model in size, potentially transforming the landscape of open-source AI research.
The Near AI Research Hub will host the project, where hundreds of contributors will collaborate to train increasingly large models. The initial study will begin with a smaller 500-million-parameter model, and as the research evolves, contributors will have the opportunity to tackle more sophisticated models in the series. Near Protocol aims to reward members through a commercialized framework, all while maintaining privacy through encrypted trusted execution environments. This approach attempts to protect user privacy while allowing for continual progress in AI capabilities.
Token sales will help fund this program, which is expected to cost roughly $160 million. Illia Polosukhin, co-founder of Near Protocol, stressed that the project has a clear plan for sustainability: monies received will not only support the original model but will also flow into a cycle of investing in subsequent models. This method capitalizes on the crypto community’s dedication to advancing decentralized technology while incentivizing continuing development.
Alex Skidanov, an AI expert, is leading the Near Protocol team in overcoming various technological challenges to bring this concept to reality. A centralized cluster of thousands of GPUs is one option for training a model of this magnitude, but Skidanov believes a decentralized method would better match with the project’s vision. This strategy would necessitate developments in distributed computing, and recent research from AI labs suggests that such breakthroughs may be on the way.
During the conference, guest speaker Edward Snowden emphasized the significance of decentralizing AI, cautioning against the dangers of a centralized AI model dominated by a single organization. He highlighted that a decentralized AI strategy is required to maintain user privacy and the key objectives of Web 3.
Near Protocol’s 1.4 trillion-parameter AI model represents a watershed moment in both the crypto and AI industries, exhibiting a shared vision for innovation that values privacy and decentralization. As the project progresses, it may potentially establish new benchmarks for how artificial intelligence might evolve ethically in a decentralized digital ecosystem.
