“Before our eyes, a paradigmatic shift is taking place.”
General Instigation
One of artificial intelligence’s most colorful figures is set to take online a supercomputing network that (he hopes) will usher in the age of human-level artificial general intelligence (AGI).
In a statement to LiveScience, SingularityNET CEO Ben Goertzel — the guy behind Sofia the Robot — said that his company’s “multi-level cognitive computing network” will be an integral step in the evolution from AI (as we know it today) to AGI.
Using some of the most advanced hardware in the world, including multiple types of NVIDIA graphics processing units (GPUs), the firm hopes to build out a new network capable of harnessing the powering of existing systems, to move from data-heavy computing to lean machines that can essentially think and reason for themselves.
This global network of supercomputers will furthermore provide the computing power necessary to host such a huge transitional shift.
Coming Soon
SingularityNET representatives told LiveScience that the first of these supercomputers will come online in September, and will continue to be brought to life throughout the end of 2024 and into 2025 (depending on when their components are delivered).
To manage this seemingly massive endeavor, SingularityNET is also building out its own software that oversees this “federated compute cluster,” which will allow for the simplification of data and protects its most sensitive aspects from exposure.
As TechSpot notes in its write-up of the new SingularityNET system, these purportedly punched-up supercomputers could be used for everything from scientific research to national security — the same tasks supercomputers are already used for, but with the addition of advanced AI and, in future, AGI.
“This supercomputer in itself will be a breakthrough in the transition to AGI,” Goertzel boasted to the website. “Before our eyes, a paradigmatic shift is taking place.”
More on AGI: OpenAI Researcher Says He Quit When He Realized the Upsetting Truth
Source link
lol
