Microsoft’s in-house AI model to compete with Google, OpenAI: What we know

Microsoft has been reportedly training a new in-house AI language model to compete with Alphabet’s Google and OpenAI.

The model is internally referred to as MAI-1 and has been the major focus of Mustafa Suleyman- Google DeepMind co-founder and former CEO of AI startup Inflection- who was recently hired by Microsoft, the Information reported.

What is the purpose of the AI model that Microsoft is preparing?
The report- which cited two Microsoft employees in the know- claimed that the exact purpose of the model has not been determined yet and will depend on how well it performs.

What we know about Microsoft’s AI model so far?
Microsoft has been setting aside servers equipped with Nvidia’s graphic processing units along with data to improve the model. MAI-1 will have roughly 500 billion parameters, the report claimed. In comparison OpenAI’s GPT-4 is reported to have one trillion parameters and Phi-3 mini measures 3.8 billion parameters.

When will the AI model be previewed by Microsoft?
The company could preview the new model as soon as its Build developer conference later this month, the report said.

How larger is the AI model and how much does it cost?
The AI model will be “far larger” than the open source models Microsoft had previously trained. This also means it will be more expensive, as per the report. It comes after Microsoft launched a smaller artificial intelligence model called Phi-3-mini last month.

Microsoft focussing more and more on AI
Microsoft has invested billions of dollars in OpenAI and deployed the ChatGPT maker’s technology across its suite of productivity software.