Nvidia CEO says he's 'empathetic' to DLSS 5 concerns — Jensen Huang doubles down on defense while decrying 'AI slop'
More of the same talking points, but it seems Nvidia has seen the backlash.
Get Tom's Hardware's best news and in-depth reviews, straight to your inbox.
You are now subscribed
Your newsletter sign-up was successful
Nvidia CEO Jensen Huang is conceding to backlash, following comments the executive made at GTC 2026 where he said gamers were "completely wrong" about their criticisms of DLSS 5, as first reported by Tom's Hardware. During a recent appearance on the Lex Fridman podcast, Huang seemed more sympathetic to the vocal crowd that has framed DLSS 5 as "AI slop."
"I think their perspective makes sense," Huang said. "And I could see where they're coming from because I don't love AI slop myself. You know, all of the AI-generated content increasingly looks similar, and they're all beautiful... so I'm empathic toward what they're thinking. That's just not what DLSS 5 is trying to do."
Although Huang is striking a more conciliatory tone, much of his response is similar to what we heard at GTC. "The artist determines the geometry, we are completely truthful to the geometry... so every single frame, it enhances, but it doesn't change anything."
Article continues belowThere was some confusion about how DLSS 5 worked when it was first announced, and although the inner workings of it still aren't clear on a technical level, Huang has said that it isn't a general-purpose generative AI model. He describes it as "content-controlled generative AI." On the other end of the spectrum, Huang also said that it isn't a post-processing filter. The technical details of DLSS 5 live somewhere between that space, and we likely won't know them until later this year when the feature is set to release.
"The question about enhancing, DLSS 5... in the future, you could even prompt it. You know, I want it to be a toon shader. I want it to look like this, kind of. You could even give it an example and it would generate in the style of that, all consistent with the artistry, the style, the intent of the artist," Huang continued. "All of that is done for the artist so they can create something that is more beautiful but still in the style that they want."
Although the talking points about DLSS 5 remain unchanged, it seems that Huang has at least heard the criticism. "I think that they got the impression that the games are going to come out the way the games are... and then we're going to post-process it. That's not what DLSS is intended to do."
Huang also made assertions that DLSS is "integrated" with the artist, and suggested that it would put the power of generative AI in the hands of artists working in game development; although, we've already seen generative AI show up in shipping releases, from Clair Obscur: Expedition 33 to the recently-launched Crimson Desert. Up to this point, DLSS hasn't been a tool developers have much interaction with. It's not post-processing, either, but it comes late in the rendering chain and is largely governed by Nvidia's models and various DLSS presets.
Get Tom's Hardware's best news and in-depth reviews, straight to your inbox.
Although DLSS 5 looks like it's doing a lot, Huang said that it's just another tool, not an essential feature. "The gamers might also appreciate that, in the last couple of years, we introduced skin shaders to game developers, and many of those games have skin shaders that include sub-surface scattering that makes skin look more skin-like... [DLSS 5] is just one more tool. They can decide what to use," Huang ended the conversation about DLSS 5. Immediately after, without missing a beat, he said 1993's Doom was the most influential video game ever made.
Follow Tom's Hardware on Google News, or add us as a preferred source, to get our latest news, analysis, & reviews in your feeds.

Jake Roach is the Senior CPU Analyst at Tom’s Hardware, writing reviews, news, and features about the latest consumer and workstation processors.
-
usertests Best response would be to keep quiet and work on the feature. Possibly scrap it, change how it works, and delay release if it wasn't actually based on "geometry" or whatever.Reply
I don't know how well regarded FSR 4.1 is vs. DLSS 4.5, but only Nvidia is likely to make Nvidia lose market share these days. ;) AMD isn't going to be rushing to introduce a DLSS 5 style capability that everyone is now wary of. -
CelicaGT Of course it's not altering geometry because it's processing a 2D image. Jensen is gone full market-speak here and it's sickening regardless of DLSS 5's merits. It is post processing but technically not because it's still in the render pipeline (as I understand it). Doublespeak, like a lawyer he speaks out of two sides of his mouth. The end result is that it does alter the image. But he can totally say it doesn't change the geometry, because it can't.Reply
*FWIW I'm in the wait and see camp at this point, I do see some promise but the way this was presented using low settings on the native render etc. is dishonest at best and Jensen jumping to defend it so vehemently and before thinking shows his true colours. He was always a bit cringe but now just comes across as slimy. -
usertests This wccftech editorial argues that Nvidia messed up by choosing the Grace thumbnail when more subtle alterations of her face were available: https://wccftech.com/nvidia-dlss-5-reveal-mistakes-op-ed/Reply
https://cdn.wccftech.com/wp-content/uploads/2026/03/nvidia-dlss-5-geforce-rtx-resident-evil-requiem-comparison-002-off-1-scaled.jpeghttps://cdn.wccftech.com/wp-content/uploads/2026/03/nvidia-dlss-5-geforce-rtx-resident-evil-requiem-comparison-002-on-2-scaled.jpeg
Whereas that infamous thumbnail was rightfully criticized by all, this other, far less discussed comparison demonstrates how the technology can largely preserve a game character's original facial structure while still greatly improving its lighting and shading. The image choice suggests that someone at NVIDIA wanted to maximize the wow factor with that specific thumbnail choice, and in doing so, essentially condemned DLSS 5 to a fierce backlash.
The writer basically reached the same conclusion that I did above:
I would even recommend delaying the actual release of the technology beyond Fall, if at all necessary. I doubt there's any risk of anyone getting a similar neural rendering tech out before NVIDIA, and it's important that the next time DLSS 5 is shown, most of the rational critiques (not the ideological, preconceived ones against any form of AI) are addressed properly.
Future versions are likely to be more deeply integrated into 3d engines:
DLSS 5 doesn't integrate with a game engine yet (which was one of the side criticisms), but that certainly sounds like the next step for the technology and would ensure better harmony with a game's original style.
-
TerryLaze Reply
The whole point is to change the image, that's the same as what dlss upscaling and frame generation does.CelicaGT said:The end result is that it does alter the image. But he can totally say it doesn't change the geometry, because it can't.
This is just applying better, or at least different, lighting instead of increasing resolution that didn't exist or create whole frames that never existed.
Wouldn't have mattered because that image still makes her lips seem fuller and redder, which was the one big complain.usertests said:This wccftech editorial argues that Nvidia messed up by choosing the Grace thumbnail when more subtle alterations of her face were available: -
CrazyCarrot911 For Flightsims it's only useful without gauges and HuD because it messes them up.Reply
As soon as you want to read your VSI you know what I mean....
Fake frames cause real problems -
hotaru251 ReplyYou know, all of the AI-generated content increasingly looks similar, and they're all beautiful... so I'm empathic toward what they're thinking.
states hates ai slop then says its all beautiful.
really wish people would stop trying to save face and just say their real thoguths outright.
also again we have conflicting statements from Nvidia about how much control artists and devs have. and i'd more likely trust the dev who did the Q&A over Jensen ,leather jacket frick gamers, Huang.
Let alone Jensen says its not generative ai..its controlled gen ai...which is gen ai still. -
PSUpower I know i'm in the minority on this one, but i'm really excited about DLSS 5.0 and its potential impact on gaming.Reply
I don't particularly like Jensen these days, but if he somehow manages to deliver even half of the things we saw on the demos, then we'll be talking about a revolution in 3d graphics.
However, DLSS 5.0 still needs a lot of refining, before it hits acceptable performance.
After all, let's not forget that, 99.99% of gamers out there , can't afford two 5090s. -
-Fran- The reason they're not making DLSS5 work in the render pipeline as you'd normally do with most tech aimed to create predictable things, it's because they don't care about it and just want to justify and showcase inference and speed on consumer cards for the AI stuff. Because they're dedicating more and more of the die area for it, I'm sure.Reply
With all the money in the world, you'd think they'd at least have the decency to use the "generation" part on the tooling side instead, for developers. Just make different die configs for AI and regular CPUs. The intersection of them is now long past.
The "classic" GPU is now in the same crossroad as the CPU was in the late 80s when accelerators were starting to make sense.
Regards. -
TerryLaze Reply
It's not generative (making random stuff up) it's deterministic (making always the same specific thing up)hotaru251 said:Let alone Jensen says its not generative ai..its controlled gen ai...which is gen ai still.
Sure for the end user they might not care that there is a difference, but there is one. -
thesyndrome ReplyHuang also made assertions that DLSS is "integrated" with the artist, and suggested that it would put the power of generative AI in the hands of artists working in game development
That's a very interesting statement, because apparently the developers of Resident Evil 9 had ZERO knowledge of DLSS 5 before the announcement, and were as surprised as the fans to see their game being used as one of the examples on stage:
PCGamer article here
So how can DLSS 5 be "integrated with the artist" if apparently the artists DIDN'T EVEN KNOW ABOUT IT? Apparently this was all agreed by suits at Capcom and Nvidia with zero input whatsoever from the developers.