3 weeks ago
10 ARTICLE AD
Update 06/06/24 at 1:43 p.m. ET: This post was updated to include additional information from Nvidia.
Nvidia’s Almost Worth 2 Trillion. Is It a Bubble, or Is It Just Getting Started?
Soon enough, anybody with an Nvidia graphics card is going to have access to a chatbot they can access from any game. “Project Game Assist” has the possibility to democratize the laborious task of optimizing your in-game graphics. At the same time, the chatbot could be taking a bite out of the folks who take the time to write guides and tips for your favorite games.
The name for the new AI program stems from an April Fool’s joke Nvidia made seven years ago. It centered around a fake G-Assist dongle that would actually play your games for you. This new, real version from the company (which has grown rich off AI) is a different sort of AI. Instead of an AI shooting your in-game enemies like the worst kind of aimbot, it will optimize your PC’s graphics settings without digging through every game’s messy options menus. As for playing the game for you, the chatbot also has the ability to offer advice on the latest meta or crafting guides if you really don’t want to bother experimenting with the game’s systems yourself.
Image: Nvidia
Gizmodo first saw hints of the company’s plans for a chatbot this year at CES. Then, it was little more than a ChatGPT clone that could supposedly work on-device for those gamers with one of the more recent GPUs. That was also where Nvidia, alongside tech startup Convai, showed off AI NPCs that could supposedly respond to users in real time. We found them pretty unconvincing, especially if the plan is to start sticking them in games for players to interact with. AI avatars are still on the docket, and now Nvidia has opened up access to these NPCs through the cloud for developers and even non-game-related companies.
G-Assist should be able to look at the detailed specs recorded by Nvidia’s hardware and tell you how to maximize your in-game performance. In a video, the GPU maker offered a few use cases, including asking the AI how best to optimize their settings to make the best use of their gaming monitor, graphics card, or CPU. In another example, the AI could generate a graph for a player’s PC latency when playing in Cyberpunk 2077.
The AI can also change your settings for you without the player needing to dig through menus. In one example, Nvidia showed how the AI could optimize power consumption by working to limit a game to a maximum of 60 FPS. The GPU maker claimed the AI could safely overclock a player’s GPU without them necessarily needing to know anything about system latency or power discrepancies.
It’s not perfect, the company told Gizmodo the settings changes still might require you to restart the game. The AI is simply adjusting settings in the game’s config file, which might require you to force a reset. Still, it’s a handy tool nonetheless, especially if you truly hate finagling every shadow or texture setting to try and reach a minimum 60 FPS.
AI that has access to your PC settings would be a boon for anybody struggling to get the most from their computer. On the other end, G-Assist will also offer you in-game advice. For example, you could ask the AI for the best early-game weapon to craft in Ark: Survival. The AI will also have the ability to read what’s happening on your screen. So, the AI could offer you suggestions if you can’t be bothered with figuring out where to drop that next skill point for your street samurai build in Cyberpunk 2077.
Nvidia told Gizmodo their AI is “a foundational pretrained model,” meaning its a standard AI that also received extra training from content sourced from Ark developer Studio Wildcard as well as the official Ark wiki. The system then uses frame capture like the existing Nvidia ShadowPlay to let the AI analyze frames for content. It can point out different types of dinosaurs and menus, then reference those back to its training data, namely all the guides that exist on Ark’s site and wiki.
We can ignore the subset of gamers stammering their usual “git gud” motto at anyone who simply needs some basic advice. Game guides have been around since the very idea of video games was barely a twinkling in Tennis for Two inventor William Higinbotham’s eye. The problem has more to do with the routine question of where AI companies get their training data from. In this case, we know that Nvidia’s AI links players to the official Ark Community Wiki, which is a player-maintained database.
Where exactly do you think Nvidia is going to get all its game advice from? Game tip blogs regularly write about weapon tier lists and guides for certain parts of games. These sites regularly depend on search traffic from stuck gamers. Nvidia’s new in-game chatbot could easily take away by offering gamers an easy way to get their answer immediately.
