NVIDIA’s attempt to revolutionize gaming with its new AI feature, Project G-Assist, has hit a rough patch right out of the gate.
Released as a part of a comprehensive update to the NVIDIA App, the feature aimed to roll out new DLSS overrides, revamped control panel settings, and enhanced MFG support for select games. Yet, the initial reception wasn’t what the tech giant hoped for. The negative feedback was swift and widespread, echoing through various online forums such as Reddit, where gamers voiced concerns over the steep hardware requirements and functionality of the AI tool.
Reflecting on the lukewarm reception, it’s difficult not to recall the rocky road NVIDIA has been on following the launch of the RTX 5000 GPUs in recent months.
The hope was that Project G-Assist would offer gamers a robust in-game assistant, capable of managing real-time diagnostics, enhancing power efficiency, optimizing game settings, or even tweaking GPU performance. All these tasks were touted as being powered by what NVIDIA calls a Small Language Model (SLM) running directly on RTX hardware. It promised to interpret natural language commands — currently supporting English — and manage everything from lighting adjustments to fan speed modifications on compatible setups.
However, the reality many users are facing is less than stellar. Even with high hopes, the actuality of using G-Assist seems to put more strain than aid on the system, evident especially given its demanding system requirements. You’re looking at the need for an NVIDIA RTX 30, 40, or 50 series GPU with no less than 12GB of VRAM, a significant demand that excludes many potential users.
Some commenters have pointed out the irony given NVIDIA’s comments that VRAM requirements would diminish over time with the help of AI. Rumors about upcoming GPUs launching with lower VRAM further complicate NVIDIA’s AI-driven future.
Peeking into the technical side, G-Assist is powered by a condensed language model based on vast datasets. What should run seamlessly on GeForce RTX hardware instead induces noticeable lag and performance drops when engaged during gaming — an issue particularly problematic when your gaming experience is front and center.
Now, let’s talk real-world experiences. Despite a few users reporting successful adjustments like GPU undervolting, the vast majority, including my colleague Richard Devine with his RTX 5080-powered setup, found the assistant far from usable. Whether running Call of Duty: Black Ops 6, Atomfall, or Avowed, interacting with the AI noticeably tanked frame rates and, at times, froze systems altogether.
The notion of an assistant meant to enhance gaming that struggles to function while games are running is perplexing, to say the least. Especially when simple queries like checking G-Sync status manage to destabilize the system.
While NVIDIA suggests that G-Assist will evolve and improve through usage, convincing players to routinely use a currently unstable tool will be a challenge.
On a brighter note, uninstalling Project G-Assist is straightforward, posing no risk to other NVIDIA app functionalities. Simply access the app, find the G-Assist section, click on the three dots in the corner, and choose ‘Uninstall.’
Have you had a chance to test out G-Assist? What’s been your experience, and what kind of hardware are you running? I’m curious to hear how it’s performing for others, so don’t hesitate to share your experience in the comments below!
