Meta today introduced Muse Spark, their first public speaking model developed under the Meta Superintelligence Labs team. Meta aims for their artificial intelligence model to be able to help anyone, at any time, on a variety of things that are important to them.
Meta says that the Muse Spark model has been developed over the past nine months, and it also comes with reasoning support for a variety of complex questions including science, math and health. Muse Spark will also power Meta AI.
Starting today, Meta will start offering updates to Meta AI with the Muse Spark model. In addition to the use of the AI model, Meta AI will also receive a cleaner interface update than before.
Muse Spark offers multi-modal support, allowing it to understand what you see (through images or videos). This also allows this AI model to be a better assistant when Muse Spark is integrated into Meta's smart glasses. At the same time, Muse Spark can also be used for website development, or mini-games easily.
Meta AI will also be able to provide better feedback, including understanding the context of the discussion, current viral topics, or providing feedback based on the various shares and discussions that are happening across Meta’s social platforms.
Meta says they will introduce this new Meta AI experience powered by Muse Spark in the United States in stages. In the coming weeks, it is expected to be offered to users globally, in addition to other Meta apps such as Instagram, WhatsApp, Facebook and Messenger.
Moving forward, Meta also says it will include Reels, and photos from social media shares directly in Meta AI responses. Meta is also offering limited access to this model via an API, and intends to offer it in open source form in the future.
