AI startup SingularityNet is set to deploy “multi-level cognitive computing network” in the coming months. It is designed to host and train the models that will form the basis of an artificial general intelligence (AGI) capable of matching — even potentially exceeding — human cognition, the company announced on Monday.
Achieving AGI is widely viewed as the next major milestone in artificial intelligence development. While today’s cutting-edge models like GPT-4o and Gemini 1.5 Pro are immensely powerful and can perform specific tasks at superhuman levels, they’re incapable of applying those skills across disciplines. AGI, though still theoretical at this point, would be free of those limitations, and able to reason and learn on its own, regardless of the task.
SingularityNet is working to build the compute base necessary to train and deploy such a system using some of the most advanced components currently on the market. Per a report from LiveScience, the startup’s modular supercomputer will sport Nvidia L40S GPUs, AMD Instinct and Genoa processors, Tenstorrent Wormhole server racks running Nvidia H200 GPUs, as well as Nvidia’s 1,500W-plus GB200 Blackwell systems.
Supercomputer architectures differ from your conventional desktop setup in that they run multiple sets of processors (both CPUs and GPUs) assembled into individual nodes. Those nodes then get daisy-chained together by the tens of thousands into the larger arrays of the overarching supercomputer.
“This supercomputer in itself will be a breakthrough in the transition to AGI. While the novel neural-symbolic AI approaches developed by the SingularityNET AI team decrease the need for data, processing and energy somewhat relative to standard deep neural nets, we still need significant supercomputing facilities,” SingularityNet CEO Ben Goertzel told LiveScience. “The mission of the computing machine we are creating is to ensure a phase transition from learning on big data and subsequent reproduction of contexts from the semantic memory of the neural network to non-imitative machine thinking based on multi-step reasoning algorithms and dynamic world modeling.”
“Before our eyes, a paradigmatic shift is taking place towards continuous learning, seamless generalisation and reflexive AI self-modification,” he continued.
The company plans to Grant public access to the supercomputer, once it comes fully online in late 2024/early 2025, using a token system. Users will purchase tokens, as they would in an old-school arcade, and then spend those tokens to get a certain number of opportunities to play with the system. The data generated by those interactions is then fed back into the system for further AGI experimentation and development.
SingularityNet is far from the only company racing to build and deploy the first AGI. The pursuit of such systems is one of OpenAI’s founding tenets while Meta’s Mark Zuckerberg has earmarked more than $10 billion for his company’s AGI R&D.