Last week, Nvidia publicly revealed DLSS 5 – and its “generative AI” to enhance the brightness of gaming scenes – which was widely condemned by the gaming community. However, in a podcast published Monday, Nvidia CEO Jensen Huang tried to distinguish the technology’s alternative, artist-directed graphical enhancements from “AI slop,” which Huang says he’s not a fan of.
As part of a nearly two-hour-long interview with the Lex Friedman Podcast, Huang was asked to explain the “drama” around DLSS5 and “gamers online.” [that] We were worried that it would make the game as slow as the AI.” Huang responded that he could “see where they’re coming from, because I don’t like the sloppiness of the AI… All AI-generated content looks increasingly the same and they’re all beautiful, so… I’m sympathetic to what they’re thinking.”
At the same time, Huang said that DLSS5 is definitely different from that kind of “slope”, because it is “3D conditioned, 3D guided.” The artists behind the game are still creating the in-game structural geometry and textures that form the “ground truth structure” on which DLSS 5 works, Huang said. “And so every single frame, it enhances but it doesn’t change anything,” he said.
However, for the most part, gamers aren’t worried about DLSS5, which is creating trippy new content from the ground up like some generative AI world models. Instead, the concern is that DLSS5’s visual “enhancements” could smooth out many disparate games toward a single, flattened, homogeneous photo-realism standard.
This is a misunderstanding of how DLSS5 works, Huang said. It’s not a technology where a game is shipped to a state and “then we’ll post-process it,” he said. Instead, DLSS5 is “integrated with the artist, and so it’s about giving the artist the tools of AI, the tools of generative AI.”
<a href