Nvidia CEO Jensen Huang: 'Human” will be the new programming language in the era of generative AI

Webmaster’s Home (ChinaZ.com) News on August 10: At this year’s SIGGRAPH 2023 conference in Los Angeles, NVIDIA released a series of announcements, highlighting The company's innovations are particularly in the areas of generative artificial intelligence (AI) and graphics.

Huang Renxun said that Nvidia made an important business decision in 2018 that will redefine the company’s future and help an evolving industry that few people realize

Huang Renxun delivered a keynote speech at the SIGGRAPH conference in Los Angeles, pointing out that although great success has been achieved, this is only the beginning of an AI-driven future. He mentioned a turning point five years ago when he chose to adopt image processing methods based on AI technology, namely ray tracing and intelligent upscaling, corresponding to RTX and DLSS respectively

He said: "We realized that raster rendering has reached its limit." He was referring to the traditional and widely used 3D scene rendering method. "2018 is the time to 'bet the whole company'. We need to reinvent hardware, software, and algorithms. And when we change CG through AI, we are also changing AI through GPU." While widespread adoption of ray tracing and DLSS in consumer GPUs and gaming is still a work in progress, the architecture they

created was found to be a perfect fit for the growing machine learning development community

. To some extent, it can be said that AI development is constrained by the availability of computing resources, so training increasingly larger generative models is best accomplished using a system like the H100 that is designed from the beginning to operate at scale. A lot of calculations

NVIDIA has been through a period similar to the Beanie Babies boom and has successfully sold a large number of products to match its server and workstation production capacity

However, Huang Renxun firmly believes that this is just a starting point. New models not only need to be trained, but also need to be run regularly by millions or even billions of users in a real-time manner to support inference computing

"Almost everything in the future will be based on LLM: 'human' will be the new programming language."

He said. Everything from visual effects to rapidly digitizing manufacturing markets, factory design and heavy industry will all adopt natural language interfaces to some extent, Huang speculates.

The above is the detailed content of Nvidia CEO Jensen Huang: 'Human” will be the new programming language in the era of generative AI. For more information, please follow other related articles on the PHP Chinese website!