- Grass, an application that enables users to monetize their Internet connections, has devised an innovative strategy amidst the increasing interest in artificial intelligence within the cryptocurrency realm. The plan involves establishing a layer 2 blockchain network on Solana to facilitate the training of AI algorithms.
- The process of training AI involves sifting through vast datasets to identify patterns, particularly in the case of extensive language models. Certain restrictions imposed by websites on data center access hinder AI training by restricting the use of these centers for web scraping activities.
- To address this challenge, Grass leverages its network of individuals who volunteer their resources to support AI training, thereby circumventing these limitations. Participants contribute their network capabilities for free and receive rewards in return.
- Founded in 2022 and headquartered in Toronto, Canada, Grass secured $3.5 million in funding in December from investors, including Polychain Capital. While the company boasts over 1 million users and clients, their identities have not been disclosed.
- Chris Nguyen, the company’s Chief Technology Officer, emphasized the significance of selecting Solana for its scalability and cost-effectiveness in handling the substantial volume of data processed by Grass. He highlighted the advantages of utilizing blockchain technology to store training data, promoting open-source AI practices and combating data manipulation that undermines AI performance.
- Nguyen underscored the transparency facilitated by blockchain in rewarding users for their contributions to the database. Despite the expenses associated with compiling AI training data, the costs escalate further when factoring in cryptocurrency transaction fees.
- Grass aims to leverage private internet connections in its decentralized system, with customers installing the Grass application assisting in AI training by routing web requests through their connections.
- The decision to allow clients access to individuals’ internet connections through Grass raises privacy and security concerns. Nguyen clarified that the application solely routes online requests through the user’s internet connection without accessing their device or browsing history.
- Grass commits to utilizing only publicly available data and refrains from scraping sites requiring login credentials. The company plans to record metadata through its layer 2 network each time data is extracted from the web, enabling transparency regarding the data sources used for AI training.
- Grass envisions processing millions of internet requests per second through its app. To avoid exorbitant costs and network congestion on Solana, the data will be transmitted to a layer 2 blockchain using zero-knowledge proofs to maintain data privacy and security.
- Companies leveraging Grass’ network for AI training purposes include a data repository utilizing forum and blog post data to train large-scale models. Despite the versatility offered by the network, potential misuse concerns have been raised, prompting Grass to implement stringent vetting processes for clients.
- Grass anticipates a broad range of applications enabled by its network, emphasizing the accessibility of public data for diverse product development opportunities. Tim Craig, DL News’ DeFi Correspondent based in Edinburgh, invites tips and inquiries at [email protected].
Visited 4 times, 1 visit(s) today
Tags: AI, Education Last modified: March 22, 2024
