Nvidia’s AI Graphics Meet The Uncanny Valley – Are Companies Building Features Nobody Wants?

When Nvidia unveiled its latest DLSS 5 technology, the goal was clear – push video game realism further than ever before using generative AI. And if that’s the main objection, I guess they’ve succeeded. Hyper-detailed faces, smoother animations and a leap toward photorealism that blurs the line between game and reality.

It was supposed to turn the industry on its head, but instead of universal applause, the internet responded with an awkward silence and mild (and not so mild) mocking.

Across gaming communities, DLSS 5 has quickly become synonymous with “yassified” characters (as the kids say these days) and uncanny, overly polished faces.

According to reports from Euronews, gamers have been poking fun at the technology, with many arguing that the visuals feel less immersive, not more, as they were intended to be.

But the thing is, the issue isn’t that the graphics are bad, quite the opposite. It’s that they’re too real, but in all the wrong ways.

The Uncanny Valley Problem

DLSS 5 leans heavily on generative AI to enhance frames and create photorealistic detail, but this is where things get complicated. The closer graphics get to reality, the easier it is for small imperfections to stand out – a phenomenon that’s become known as the uncanny valley.

Instead of immersion, players are actually starting to experience discomfort. Faces look almost human, but just not quite. Movements feel slightly off, and rather than enhancing gameplay, the technology risks pulling players out of it entirely.

Now, this isn’t a new problem in gaming or animation, but DLSS 5 is pushing it into the mainstream.

Sometimes, Technical Progress Misses the Point

For some, the backlash raises a deeper question about priorities in AI development.

David Sherman, Head of Brand Strategy, puts it bluntly:“While the gaming world debates whether Nvidia’s new AI graphics are “AI slop,” there’s a bigger question: why is the world’s leading GPU company investing in cosmetic features when we’re in the middle of a compute crisis?

“Research teams wait months for GPU access. AI startups can’t afford the infrastructure they need. Universities ration compute time. Meanwhile, Nvidia is directing R&D toward making video game characters look like they’re ready for a photoshoot.

“This mirrors the hyperscaler problem in cloud computing. When a handful of companies control infrastructure, they decide what gets built and who gets priority. Everyone else waits in line. The demand for accessible, affordable compute has never been louder. Meeting that with shinier graphics instead of more supply is another symptom of the unhealthy monopoly in the AI.”

It’s a sharp critique, and one that taps into a broader tension across the AI industry. Indeed, as demand for compute skyrockets, decisions about where resources are allocated are becoming increasingly scrutinised.

The Question They Forgot To Ask: Do Gamers Even Want This?

The backlash isn’t just anecdotal; it’s actually backed by data.

According to insights from GWI, sentiment around AI-generated content in gaming is mixed at best:

• 87% of gamers say that knowing an artwork is made by a human makes it more meaningful
• 61% believe art should express a deeper meaning
• 39% don’t think AI is capable of producing “real” art
• 41% aren’t interested in engaging with AI-generated art

Interestingly, there’s still nuance here. Indeed, 37% of gamers do actually believe AI can create real art. But overall, the data suggests hesitation rather than excitement.

Style Over Realism

For many in the industry, the issue isn’t just about AI – rather,  it’s about misunderstanding what makes games successful in the first place.

Carl Kuhn, Executive Gaming Director and Member of the Board at JvM SPORTS, explains that, “whilst it’s objectively impressive, Nvidia’s DLSS5 highlights a familiar issue: just because you can push realism further doesn’t mean you should – or that people actually want it.

“The most successful games today aren’t winning because they look real – they’re winning because they look distinct. Clair Obscur didn’t win Game of the Year for its realistic depiction of skin folds, but for its uncompromising artistic vision. And blocky titles like Minecraft haven’t built massive audiences despite their stylisation, but because of it.

“Strong art direction beats technical fidelity every time. Differentiation doesn’t come from polishing pixels – it comes from a clear visual point of view. Hyperrealism, ironically, often feels generic.”

His point cuts to the core of the debate – realism isn’t the end goal of gaming. Escapism is.

A Bigger Question for AI

DLSS 5 may still find its place – we’re not going to count it just yet. The technology itself is undeniably powerful, and its applications could extend far beyond gaming into film, simulation and virtual production.

But, the reaction highlights something bigger, and that’s what we should be focused on.

In a world where AI can build almost anything, the challenge is no longer can we? Rather, the question is, should we? We need to start focusing on not whether we hae the technical ability to create or produce something and rather whether people even want it in the first place.

For startups and tech giants alike, this is becoming a defining question. Because, as Nvidia is learning in real time, building more advanced features doesn’t guarantee users will want them.

And in some cases, pushing too far toward perfection might be exactly what breaks the illusion.