(Credit: PCMag)
At first glance, some gamers may hate Nvidia’s DLSS 5 and its use of AI to add ultrarealistic effects to PC games. But the company’s CEO, Jensen Huang, says the early critics are “wrong.”
“We created the technology, we didn’t create the art,” Huang said, describing DLSS 5 as merely a tool that game developers can choose to use.
In a Q&A with journalists at the company’s GTC event, Huang was asked about the initial harsh reception to DLSS 5, which uses a new AI “neural rendering” model to add photorealism to game characters, objects, and environments.
The company has described it as the biggest leap in gaming graphics in years. However, some consumers have mockingly compared DLSS 5 to the AI-powered face filters available in some social media apps. Others, including IGN (a sibling site of PCMag, also owned by our parent company Ziff-Davis), have argued the technology represents a “slap in the face” to the art of video game design.
Huang: 'They're Completely Wrong'
In the Q&A, Huang was specifically asked about how DLSS 5 allegedly creates “worse, homogenous” imagery or imposes Nvidia’s view on how games should look. In response, Nvidia’s CEO said, “First of all, they’re completely wrong.”
Huang then alluded to how DLSS 5 was designed to understand 3D characters, environments, and their motion to add a new level of realism, as opposed to ham-fisted AI face filters.
(Credit: PCMag)“As I explain very carefully, DLSS 5 infuses controllability of geometry and textures and everything about the game with generative AI,” he said. "Now, you can still fine-tune the generative AI so that you could make it your artistic style. And so all of that is up to you. We created the technology; we didn’t create the art."
“DLSS 5, because it's geometry-controlled, it’s conditioned by the ground truth of the game," he added. "It enhances and adds generative capability to it, but it doesn't change artistic control. It’s not post-processing at the frame level, it’s generative control at the geometry level.”
(Credit: PCMag)“This is very different from generative AI,” he later elaborated further. “It’s content-controlled generative AI. That’s why we call it neural rendering.”
Nvidia emphasized similar points during our hands-on with DLSS 5. Although the technology functions as a proprietary model, the company says game developers will be able to fine-tune the neural rendering effect to their liking.
Still, some gamers, including PCMag staff members, say that DLSS 5 seems to go overboard with AI-enhanced effects, creating hyper-real faces that give "uncanny valley" vibes. For now, Nvidia noted in a FAQ: “DLSS 5 at GTC is an early preview, and the model is still being optimized. We will share these details closer to release in fall 2026.”
During his Q&A, Huang also mentioned ray tracing, which Nvidia introduced to GeForce RTX 2000 GPUs back in 2018. He noted that the enhanced lighting and shadow effects initially faced criticism, too, but the technology improved and became mainstream in gaming over time.
“Everybody pooh-poohed it," he said. "Everybody said ray tracing was FUBAR. If we didn’t have RTX today, doing full scene path-tracing, computer graphics wouldn’t be what it is today.”