Microsoft launches first self-evolving AI model

According to Microsoft, the MAI-Voice-1 model is so powerful that it can generate 1 minute of sound in less than 1 second with just one graphics processing unit (GPU).

Microsoft on August 29 announced the first two Artificial Intelligence (AI) models developed and trained internally by the corporation, marking an important turning point in its strategy to reduce dependence on external technology.

The first product, called MAI-Voice-1, focuses on natural voice generation.

According to Microsoft, this model has outstanding capabilities when using only one graphics processing unit (GPU) it can create 1 minute of sound in less than 1 second.

MAI-Voice-1 is now integrated into several Copilot services, such as Copilot Daily - which provides daily audio news, or podcast talk shows to explain topics.

Users can also test the model via the Copilot Labs platform, with the option to adjust the tone and style of presentation.

In parallel, Microsoft also introduced MAI-1-preview - a foundational language model trained entirely in-house. The training process used up to 15,000 Nvidia H100 chips, allowing the model to process written instructions and provide useful answers for everyday needs.

According to Microsoft, MAI-1-preview is just a "trial," providing a look at what Microsoft will deploy in the Copilot ecosystem in the coming time. The model has begun testing on the LMArena platform, where AI systems are evaluated, and will gradually be integrated into some Copilot services in the coming weeks.

The move comes as Microsoft's Copilot services still rely heavily on OpenAI technology.

Microsoft believes that building a specialized, multifunctional AI toolkit to serve a variety of needs and purposes will bring great value to users and pave the way for a new phase of competition in the global AI race./.

(TTXVN/Vietnam+)

Source: https://www.vietnamplus.vn/microsoft-ra-mat-mo-hinh-tri-tue-nhan-tao-dau-tien-tu-phat-trien-post1058812.vnp