PC Gamer ran a poll asking its readers a direct question: “Could Nvidia convince you to use DLSS 5?” The results came back clear. Out of 4,484 total votes, a full 71% of respondents said they could not be convinced to turn the feature on under any circumstances.
Only 10% said they are already on board with the technology, while another 10% said they would consider it if it looked better than the initial showcase. The remaining 9% said they’d use it, but only in certain games.
The poll was not a simple yes or no. It included nuanced options to capture exactly how players felt, and the breakdown makes the rejection even more striking. Thirty-seven percent said they would refuse to use DLSS 5 no matter how good the final image looked, citing ethical or personal reasons.
Another 34% said it simply looked bad and that was enough to keep them away. Combined, those two groups form that 71% wall of resistance that Nvidia is now staring down ahead of a Fall 2026 launch.
What’s DLSS 5 and why did everything go wrong
To understand why the backlash hit so hard, it helps to know what DLSS 5 actually is, and how different it is from everything that came before it.
DLSS launched alongside the RTX 2080 in 2018 as a tool to help games run faster. Over the next eight years it evolved steadily, moving from resolution upscaling to full frame generation. By the time DLSS 4.5 arrived, it was drawing 23 out of every 24 pixels on screen through AI, and most players were fine with that because the games still looked like themselves.
DLSS 5 is a different animal entirely. Revealed at GTC 2026 on March 16, it does not upscale or generate frames, it rewrites them. The technology takes each frame’s color and motion vectors as input and runs them through a neural rendering model that replaces the original lighting, textures, and material properties with AI-generated alternatives. The results, according to NVIDIA, are meant to look more photorealistic. According to the internet, they look like AI slop.
The most talked-about example was Grace Ashcroft, the lead character from Resident Evil Requiem. Her face appeared visibly altered in Nvidia’s own demo, smoother skin, sharpened features, a kind of uncanny valley glow that gamers immediately recognized as AI-generated.
What made it worse was that Capcom’s artists had not been consulted or informed beforehand. The company later stated it would not use generative AI for final in-game assets, models, or textures. When the developers of your showcase game publicly distance themselves from your technology, that is not a good launch week.
Jensen Huang steps in, first to fight, then to soften
Nvidia was not going to stay quiet. CEO Jensen Huang addressed the backlash directly at a press Q&A with Tom’s Hardware during GTC 2026, and his first response did not exactly calm things down. “Well, first of all, they’re completely wrong,” Huang said when asked about the widespread criticism.
He argued that DLSS 5 is not a post-processing filter slapped on a finished image, but a system grounded in actual 3D geometry. He called it “content-control generative AI” and insisted that developers retain full artistic control over how it is applied.
Calling your customers completely wrong in public, while they are already flooding social media with memes about your product, is a calculated risk. It did not pay off.
A few days later, Huang appeared on the Lex Fridman podcast and struck a noticeably different tone. “I think their perspective makes sense,” he said. “I don’t love AI slop myself. All of the AI-generated content increasingly looks similar, and they’re all beautiful, so I’m empathetic towards what they’re thinking.”
He also hinted that DLSS 5’s open system would eventually allow developers to train their own models, steer toward specific art styles, or even prompt the AI to match a particular aesthetic.
The problem is that the messaging coming out of Nvidia has not always been consistent. While Huang described DLSS 5 as operating at a 3D level with full scene awareness, an Nvidia GeForce Evangelist confirmed separately that the model actually uses a 2D frame plus motion vectors as input, which lines up exactly with what critics feared, that DLSS 5 is essentially a very advanced 2D filter inferring what the image should look like rather than truly understanding it. That contradiction has kept the skeptics firmly in their corner.
DLSS 5 is still in early preview and is not due until Fall 2026, so Nvidia has time to refine the technology, dial back some of the heavier-handed effects, and present a version that feels less like an AI filter and more like a genuine enhancement tool. It has pulled off that kind of course correction before.
DLSS 1 was widely mocked when it launched, and by DLSS 2 most of the community had come around. Whether history repeats itself here is the real question, and based on that poll, Nvidia has a very long way to go.
What’s your take, would you give DLSS 5 a shot when it launches, or are you firmly in the 71%? Tell us in the comments!