Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
Nvidia unveiled a prototype AI avatar that lives on your computer desktop at CES 2025. AI assistant, R2Xlooks like a video game character and it can help you manage programs on your computer.
The R2X avatar is created and animated using Nvidia’s AI models, and users can run their avatar in popular LLMs of their choice, such as OpenAI’s GPT-4o or xAI’s Grok. Users can talk to R2X via text and voice, upload files to it for processing, or even let the AI assistant see what’s happening live on your screen or camera.
Tech companies have recently been creating a lot of AI avatars, not only in video games, but also for enterprise and consumer customers. Early demos are strange, but some think these avatars are a promising user interface for AI assistants. With R2X, Nvidia seeks to combine generative video game capabilities with advanced AI assistants to create an AI assistant that looks and feels human.
It’s very similar Microsoft’s Recall feature (this happened delayed due to privacy concerns), R2X can take permanent screenshots of your screen and run them through an AI model for processing, although this feature is disabled by default. When enabled, it can offer feedback about programs running on your computer and help you work through a complex coding task, for example.
R2X is still a prototype, and even Nvidia admits it still has some bugs to work out. In demos with TechCrunch, Nvidia’s avatar had an uncanny valley feel to it — its face was sometimes stuck in odd positions, and its tone seemed a bit aggressive at times. In general, I think it’s weird to have a slightly humanoid avatar looking at me while I’m doing my work.
He generally offered helpful instructions and saw exactly what was on the screen. However, at one point the avatar gave us wrong instructions and later the avatar stopped being able to look at the screen at all. This may be a problem with the underlying AI model (in this case, GPT-4o), but the example shows the limitations of this early technology.
In a demo, an Nvidia product manager showed how R2X can see apps on your screen and help users. In particular, R2X helped us use Adobe Photoshop’s generative fill feature. The photo we chose was Nvidia CEO Jensen Huang standing in an Asian restaurant with two restaurant workers. Nvidia’s avatar was hallucinating and giving false instructions on where to find the generative fill function. But after switching the AI model we used to xAI’s Grok, the avatar regained its ability to look at the screen.
In another demo, R2X was able to receive a PDG from the desktop and then answer questions about it. This process is equipped with a local search augmented generation feature that enables these AI avatars to extract information from a document and process it using its underlying LLM.
Nvidia uses some of its video game division’s artificial intelligence models to power the look of these avatars. It uses Nvidia RTX neural faces algorithm to create avatars. It uses a new model called Nvidia to automate face, lip and tongue movement Audio2Face™-3D. That model seemed to stand still at some points, holding the avatars’ faces in awkward positions.
The company also says that these R2X avatars will be able to join Microsoft Teams meetings, acting as personal assistants.
Nvidia’s product chief says the company is also looking to give these AI avatars agent abilities, so that R2X could one day perform actions on your desktop. These capabilities appear to be a long way off, and they will likely require collaboration with software makers such as Microsoft and Adobe, who are also trying to develop similar agent systems.
It’s not immediately clear how Nvidia creates sounds in these products. R2X’s voice when using GPT-4o sounds unique from any of ChatGPT’s preset voices, while xAI’s Grok chatbot does not yet have a voice mode.
The company plans to launch these avatars in the first half of 2025. Nvidia sees this as a new user interface for developers, allowing users to embed their favorite AI software products or even run these avatars natively.