Meta's new AI Model Will Make a Metaverse Avatar Groove Like A Human

Meta has said that it is releasing an artificial intelligence model called Meta Motivo, which could control the movements of a human-like digital agent, with the potential to enhance Metaverse experience.

The company has been plowing tens of billions of dollars into its investments in AI, augmented reality and other Metaverse technologies, driving up its capital expense forecast for 2024 to a record high of between $37 billion and $40 billion.

Meta has also been releasing many of its AI models for free use by developers, believing that an open approach could benefit its business by fostering the creation of better tools for its services.

"We believe this research could pave the way for fully embodied agents in the Metaverse, leading to more lifelike NPCs, democratization of character animation, and new types of immersive experiences," the company said in a statement.

Meta Motivo addresses body control problems commonly seen in digital avatars, enabling them to perform movements in a more realistic, human-like manner, the company said.

Meta said it was also introducing a different training model for language modeling called the Large Concept Model (LCM), which aims to "decouple reasoning from language representation".

"The LCM is a significant departure from a typical LLM. Rather than predicting the next token, the LCM is trained to predict the next concept or high-level idea, represented by a full sentence in a multimodal and multilingual embedding space," the company said.

Other AI tools released by Meta include the Video Seal, which embeds a hidden watermark into videos, making it invisible to the naked eye but traceable.

The development comes shortly after the company updated Meta AI such that it is now powered by the company's Llama 3.2 large language model. Llama 3.2 brings image editing and voice features for the company's artificial intelligence-based chatbot. Simput put, the update enables users to edit an existing image as opposed to just creating an image based on text-based prompts. The update also enables users to interact with Meta AI using voice-based prompts. Earlier, users could interact with Meta AI only with text-based prompts.