AI is all the rage right now, and it doesn’t matter if you are talking about movies, music, or video games, it seems as though every piece of media is being affected by it, for better or worse. In the case of video games, AI use in development has become increasingly interesting over the past few months, not just because of its place in generating art, but also due to how technologies like frame generation and Nvidia’s DLSS can fundamentally alter the experience on-screen. This most recent iteration with DLSS 5 is perhaps the most contentious, as we are now entering a truly strange territory where the visuals have become closer to reality than ever before.
However, unlike in the previous iterations where the promise was to boost your performance without trading off the graphics, DLSS 5 seems to go the opposite direction, putting a much bigger strain on the hardware and offering you a choice between the two, rather than the best of both worlds. In fairness to Nvidia, the current release that came with NBA 2K27 on PC and is being tested in every game under the sun is still in its early stages and will get better with time and as developers actually apply the technology in the correct places. Yet, the larger question remains about frames versus graphics, and with how complex the hardware market is nowadays, the worry is that we may end up in a situation where the best visuals become completely locked behind huge paywalls, and good performance once again becomes an option as opposed to being a guarantee.
DLSS 5 Pushes the Performance Versus Graphics Debate The Other Direction
Graphics in gaming have clearly evolved a lot in the past decade. While older titles like The Witcher 3 and Red Dead Redemption 2 still hold up incredibly well, there have still been some significant strides taken in the visual department. A lot of the advancement comes down to new technologies making better graphics more accessible and, more importantly, possible to produce, but I think the biggest milestone came from the introduction of Nvidia’s DLSS. With DLSS, you could achieve top-notch visuals and still keep the performance of lower graphical settings at the click of a button, and most importantly, the technology could be applied to both new and older releases, giving a lot of classics a second lease on life.
It felt like a piece of magic that you could use in a game to leverage the power of AI and make a weaker system run much heavier games than it could have ever dreamed of. By clicking the toggle in the settings, I often got close to double the frames, with no noticeable dip in visuals, and that is on 5+ year-old hardware that has no business being able to support Black Myth: Wukong and Resident Evil Requiem at more than 30fps. Each new iteration of DLSS only made things better, and as we drew closer to the infamous DLSS release, many people were excited to see what Nvidia had been cooking up behind the scenes and if those hyperrealistic visuals could actually be achieved in real conditions.
With DLSS, you could achieve top-notch visuals and still keep the performance of lower graphical settings at the click of a button…
As it turns out, the graphics were far from make-believe, and now, thanks to some clever modding, we can see how DLSS 5 looks and functions in a much broader range of titles. But it didn’t take long for people to realize that this version of the tool was very different, and not in a good way. Unlike the previous DLSS implementations, the performance hit is significantly higher, and even more so in the demanding games that already ask a lot of your hardware. You might get some pretty faces and awesome lighting in the background, but just like any other visual setting increase, the frame rates just aren’t going to keep up for long.
This harkens back to a classic ideology that has existed for so long in the gaming world, particularly on PC. In the past, because of how many variables there are in creating a game, developers could only do so much. As a result, they would need to decide whether they valued a crisp 60FPS at all times, or if they needed those 4K visuals in order for players to have a good time. DLSS changed that philosophy entirely, basically saying that players can get the best of both worlds if they want, but DLSS 5 pushes the needle far more towards the graphical side of the spectrum. It might still follow the same steps of rendering a lower-resolution image and upscaling it, yet now, the focus is far more on increasing graphical fidelity with AI than improving frame rates.
The Bigger Problem Still Exists
As great as all these advancements are, I can’t help but feel we are falling into the same trap that is currently harming the game industry from all sides. Right now, a lot of games are launching with poor performance, and the only way you can get things to a usable level is to turn on those AI upscalers and hope for the best. And if DLSS 5 is going to become the norm for achieving good visuals, then it starts to feel as though the choice of visuals or performance is far out of your hands. Getting playable frames and these realistic visuals is all well and good for someone with an RTX 5090, but lower-end systems just cannot make use of the technology, assuming they can even use DLSS to begin with.
What that means is developers, like they are already doing, may start placing a greater emphasis on using the DLSS 5 tech, showing it off in marketing and promising specific stats like 1440p visuals at X frames, but locking that standard behind a beefy paywall that is only going to get worse from here. This problem goes both ways actually, as a person who can’t afford a higher-end rig is never going to be able to get the ultra-tier graphics from the trailers, but at the same time, someone who is willing to reach deep into their wallets is going to be greeted with those stellar visuals, yet suffer greatly from a performance hit even on their spaceship of a PC.
The point is that DLSS 5 feels like one step forward and two steps back, where the existing tech was intended to deliver a one-size-fits-all solution to a global issue, but it now exists as a choice reserved for a select few that comes with the same tradeoff it was supposed to correct in the first place. Additionally, we need to consider other AI software, such as frame generation, and how they fit into this conversation about trade-offs in game design and the user experience. These technologies aren’t going anywhere any time soon and are finding their way into virtually every new release, which may very well mark the start of a new era for gaming.
We may end up in a situation where all your frames are generated, all the graphics are upscaled, and even the hardware you are playing on exists on the cloud, due to the increasing inaccessibility for the average user. In that situation, you are no longer trading off performance and graphics, but trading a genuinely well-crafted game for one that has to be bolstered and corrected by the mystical power of AI. Yet, as powerful as these tools are and will become, they are still only band-aids that can help address visual and functional problems, and if the underlying game just isn’t good, then no amount of generation or computational power is going to fix it.
