AGI breakthrough imminent as supercomputer network takes shape
SingularityNET, a leading artificial intelligence (AI) company, is set to accelerate the development of artificial general intelligence (AGI) through a global network of supercomputers. The first machine in this series is tipped to come online in September this year. The company's goal is to establish a "multi-level cognitive computing network" that will host and train the necessary architectures for AGI, according to SingularityNET representatives.
AGI: A leap beyond current AI capabilities
AGI is a futuristic concept that aims to outperform human intelligence across various disciplines and improve its decision-making through self-learning and data access. This is a significant advancement from current AI technologies like machine learning and generative AI systems, which excel in specific areas based on their training data but struggle with cognitive or reasoning tasks across different fields.
Supercomputers to integrate advanced AI systems
The supercomputers being developed by SingularityNET will incorporate elements of advanced AI systems. These include deep neural networks, large language models (LLMs), and multimodal systems that link human behaviors like speech and movement inputs with multimedia outputs. While the first supercomputer is set to go live soon, its completion is targeted for late 2024 or early 2025, depending on supplier delivery timelines.
Modular machine to feature these components
The modular supercomputer will be equipped with advanced components such as NVIDIA's L40S GPUs, AMD's Instinct and Genoa chips, Tenstorrent Wormhole server racks featuring NVIDIA H200 GPUs, and NVIDIA's GB200 Blackwell systems. These components represent some of the most powerful AI hardware available today. SingularityNET CEO Ben Goertzel has highlighted that this supercomputer will mark a significant step toward AGI.
SingularityNET's unique software for AGI development
To facilitate the growth of AI and AGI, SingularityNET has developed a unique software called OpenCog Hyperon. This open-source framework is specifically designed for the architecture of AI systems. Users will be able to access the supercomputer via a tokenized system, allowing them to use and contribute data to existing sets that other users depend on for testing and implementing AGI concepts.
