NVIDIA's CEO Jensen Huang's defence of DLSS5 (Deep Learning Super Sampling) on the Lex Fridman Podcast aims to reassure industry professionals and stakeholders about the technology's direction amid intensified backlash, fostering confidence in NVIDIA's commitment to innovation.
Jensen Huang, in his recent speech on Monday, March 23, on the Lex Fridman podcast, addressed the tensions and controversy surrounding the new Version of DLSS-5, with his statement, "I think we've achieved AGI".
It was first launched in 2019 by NVIDIA. DLSS began as an AI-powered upscaling tool designed to improve gaming performance by rendering frames at lower resolutions and then enhancing them using deep learning.
The technology was moderately well-received by the audience after the release of DLSS 2 and DLSS 3, thanks to significant improvements in image quality and frame generation. However, the release of DLSS 5 has marked a major shift in how the technology operates.
Unlike earlier versions that focused mainly on upscaling and performance optimisation, DLSS 5 introduces elements of generative AI and neural rendering, which some industry professionals see as a potential risk to artistic control and visual consistency in games.
In the podcast, he says he can feel the gamers who hate AI-generated content, and he also acknowledges that he hates it himself.
According to Huang, DLSS 5 is guided by the artistry of the artists, with its core purpose being to enhance every single frame without changing anything. Furthermore, the Nvidia boss revealed that the system is open, and developers could prompt DLSS 5 in the future to achieve a highly specific look.
The increased role of AI in scene creation, particularly concerning lighting and textures, has drawn criticism from both gamers and developers who prioritise artistic integrity. The feedback raises questions about how to strike a balance between innovation and creative authenticity while also supporting developers in various aspects.
