Jen-Hsun Huang: Five years ago, we had huge expectations for AI

Nvidia founder and CEO Jensen Huang delivered a keynote speech at the SIGGRAPH conference in Los Angeles. He noted that the company made a crucial business decision in 2018 that will redefine the Its future and promote the continuous development of the industry

IT House noticed that in his speech, Huang Jenxun recalled the critical moment five years ago when he chose to use ray tracing and intelligent rendering to process images using AI, namely RTX and DLSS technology. "We recognize that rasterization has reached its limits," he noted. Rasterization is a traditional and widely used method for rendering 3D scenes. "2018 is a decisive moment that forces us to reinvent hardware, software and algorithms. As we redefine computer graphics with AI, we are also redefining the use of GPUs in AI."

While the adoption of ray tracing and DLSS in consumer GPUs and gaming is still evolving, their architecture fits perfectly with the growing machine learning development community. In order to train larger and more complex generative models, significant computing resources are required, and systems like the H100 that are designed from the ground up to perform the necessary operations scalably can provide this resource. It can be said that, to some extent, AI development is only limited by the availability of computing resources

Huang Renxun insists that this is just the beginning, and that new models need to be run regularly in real time by millions or even billions of users, not just for training. He said that in the future, large-scale language models will be at the forefront of almost all fields, and humans will become a new programming language. Huang speculates that areas ranging from visual effects to rapidly digitizing manufacturing, factory design and heavy industry will all adopt natural language interfaces to some extent. "Entire factories will be software defined and robotic, and the cars they build will themselves be robotic," he said. "So it's robots designing robots building robots."

Huang Renxun did not touch on the challenges and regulatory issues facing AI

The above is the detailed content of Jen-Hsun Huang: Five years ago, we had huge expectations for AI. For more information, please follow other related articles on the PHP Chinese website!