Nvidia has announced that the newly developed G-Assist system assistant will officially launch in February, providing services for GeForce RTX users. This AI assistant was first showcased at the June 2024 release event and aims to help users better manage and optimize various settings on their personal computers through simple voice or text commands.
As modern PC performance continues to improve, users face thousands of hardware and software configuration combinations when setting up their PCs for optimal performance. To address this complexity, NVIDIA has introduced Project G-Assist, an experimental AI assistant that runs on GeForce RTX AI PCs. G-Assist is designed to help users control various PC settings through straightforward voice or text commands, including optimizing game and system settings, monitoring frame rates, and other key performance metrics, as well as controlling peripheral settings like lighting.
G-Assist utilizes a specially tuned Small Language Model (SLM) that efficiently understands natural language commands and calls upon various NVIDIA and third-party PC APIs to execute actions. It can provide real-time system diagnostics and suggestions, helping to alleviate system bottlenecks, improve power efficiency, optimize game settings, and even overclock the GPU.
G-Assist will be offered through a layer of the NVIDIA application, allowing users to input commands in a chat interface or hold down a push-to-talk key to converse with G-Assist. Unlike large cloud-based AI models that require online access and paid subscriptions, G-Assist runs on the user's GeForce RTX GPU, making it responsive, free to use, and capable of offline operation.
The core of G-Assist is based on an instruction model from Llama, featuring 3 billion parameters with strong language understanding capabilities, yet its size is less than 1% of current large AI models. This enables G-Assist to run smoothly on various RTX hardware. In terms of AI assistant development, NVIDIA has also introduced the ACE technology suite, which OEMs and ISVs have already utilized to develop custom AI assistants.
Additionally, NVIDIA plans to open this framework to the broader AI community, providing development tools to encourage developers to create plugins that expand G-Assist's functionality. NVIDIA will release a GitHub repository offering samples and documentation for developers to test and submit new features locally.
G-Assist will appear as a new entry in the "Discover" section of the NVIDIA application for GeForce RTX users to download and install. In the meantime, users can also follow the latest news about the GeForce RTX 50 series to learn about NVIDIA's innovations and advancements in gaming and applications.
Official blog: https://www.nvidia.com/en-us/geforce/news/g-assist-ai-companion-for-rtx-ai-pcs/
Key Points:
🎮 G-Assist will officially launch in February, aiming to simplify PC setup operations for GeForce RTX users.
💡 G-Assist uses a small language model that can understand natural language and optimize system performance.
🔧 NVIDIA will also open the G-Assist development platform, encouraging the community to develop plugins to extend its functionality.
