Microsoft is said to be investing in a new in-house AI language model called MAI-1, which is said to be large enough to compete with Google and OpenAI, though its purpose is said to be determined by its performance
Microsoft is investing in an in-house AI language model, referred to as “MAI-1”. The new model is being overseen by Mustafa Suleyman, the Google DeepMind co-founder and former CEO of AI startup Inflection, according to the Information.
Microsoft bought the intellectual rights of the AI startup at $650 million in March, and also hired its majority of the staff. That being said, the new AI language model is different from previous models released by Infection, and certainly not carried over, as per the knowledge of two Microsoft staff, reports the Information.
The new model will have around 500 billion parameters. In comparison, Microsoft’s recently released Phi-3-mini measures 3.8 billion parameters, Phi-3-small measures 7 billion parameters and Phi-3-medium 14 billion parameters.
Large language models or LLM are part of foundation models which are trained on vast amounts of data. They provide essential skills used across various applications and perform a horde of tasks.
Microsoft has invested over $10 billion dollars in OpenAI and deployed the technology across all its suite of software which in turn pushed the company in the lead in the AI race.
The new MAI-1 is expected to be large enough to compete with Google, and OpenAI, even though the model’s exact purpose has yet to be determined. The report also states the possibility of Microsoft previewing the new model at its upcoming Build developer conference.