Artificial Intelligence Hardware
A Korean research team has succeeded in developing an original semiconductor technology that can function as a human brain by assembling a semiconductor with nanomachines.
The National Research Foundation (NRF) announced on Oct. 19 that a research team headed by Choi Woo-young, professor of the Department of Electronic Engineering at Sogang University, successfully developed a tech for a 3D artificial intelligence semiconductor chip. The von Neumann method, a representative technique for semiconductor chips, can rapidly and effectively carry out simple tasks, but there is a certain limit to flexibility, adaptability, evolution, and learning, which are all qualities that are required to functioning like a human brain.
The research team conducted a study on new structures and voltage levels of nanomachines to play the role as a switch to deliver signals in electronic circuits, in order to integrate the machine into a 3D form similar to a human brain. They were able to integrate a nanomachine circuit into an existing 2D Complementary Metal-Oxide Semiconductor (CMOS) in a 3D fashion.
They identified the fact that the new semiconductor chip could function like a human brain using less than 50 percent of the energy of existing chips. The research team found that the newly-developed artificial intelligence semiconductor chip based on a 3D convergence integration tech has a 400 percent improved density and consumes less than 50 percent of the energy compared to a one using 2D-based CMOS circuit tech.
The study was funded by Korea’s Ministry of Science, ICT and Future Planning and the NRF through a support project for leading researchers, and the research findings were published online in the Sept. 1 issue of IEEE Electron Device Letters, a scientific journal published by the Institute of Electrical and Electronics Engineers (IEEE).