Nvidia just introduced a platform called Omniverse Avatar, created specifically to generate immersive avatars powered by Artificial Intelligence.
The idea is to take advantage of voice generation systems, computer vision, natural language understanding and simulation to create avatars that are as complete as possible.
These avatars will be able to communicate with human users within collaboration platforms, within simulated worlds, such as the metaverse that we talk so much about lately, or their own Nvidia Omniverse.
One of the goals is to allow you to create avatars that can talk about a wide variety of topics and conduct customer service interactions, including making personal reservations to order food at restaurants, for example. You can see examples from minute 27 of this video:
From 1 hour and 22 minutes you have the demonstration of the kiosk.
Your target audience is marketers, who will be able to create solutions to interact with your customers, intelligent virtual assistants that compete with those that can be created with other solutions, such as Deepbrain, Soul Machines and AI Foundation.
By integrating with Nvidia Omniverse, which already has 70,000 individual creators, they are not starting from scratch, so they can offer the service to companies already working with this omniverse.
They use the Megatron 530B Large Language Model to give them the ability to recognize, understand and generate human language, allowing avatars to ask questions on a wide range of topics, summarize complex stories in short formats, recognize speech in multiple languages. and then translate it.
They’re not yet available for real-world testing, but the demos they’ve done are showing promise. We are very close to starting to interact with 3D avatars in stores.