His company, he said, has had a decade head start, embarking on its first supercomputer in 2012. The machine, which was delivered five years later, changed the way computers worked, allowing them to generate sentences, images and videos. We realized that deep learning and A.I. was not a chip problem. It's a reinvention-of-computing problem, Mr. Huang said, speaking at the DealBook Summit in New York.
Mr. Huang said that it will be a decade before A.I. companies achieve so-called artificial general intelligence, or A.G.I., the state at which a machine could break down tasks the way a human brain can. Until then, he predicted that businesses will develop custom A.I. capabilities to perform a specific aim such as drug research or design semiconductors. Nvidia already relies on A.I. for its core business, he said.
[
add
]
[
|
|
...
]