Microsoft in-house AI model to rival Google and OpenAI

By Consultants Review Team Tuesday, 07 May 2024

Microsoft is purportedly training a new internal AI language model in order to take against Alphabet's Google and OpenAI. According to the Information, Mustafa Suleyman, the co-founder of Google DeepMind and the former CEO of the AI firm Inflection, has been concentrating on the model, known internally as MAI-1, since he was recently hired by Microsoft.

What is Microsoft planning to do with its AI model?

According to the story, which included two Microsoft workers who were in the know, the model's precise goal has not yet been established and will rely on how well it functions.

What is currently known about Microsoft's AI model?

Microsoft has started allocating servers that are outfitted with Nvidia GPUs in addition to data in order to enhance the model. According to the research, MAI-1 would have almost 500 billion characteristics. Comparatively, Phi-3 micro measures 3.8 billion parameters, while OpenAI's GPT-4 is said to contain one trillion parameters.

When will Microsoft release a preview of the AI model?

According to the sources, the corporation may provide a sneak peek at the new model during its Build developer conference later this month.

What is the size of the AI model and what is the cost?

Compared to the open source models Microsoft had previously trained, the AI model will be "far larger". The analysis also indicates that this will result in a higher price. It follows Microsoft's announcement last month of Phi-3-mini, a scaled-down artificial intelligence model.

Microsoft is increasingly concentrating on AI

Microsoft has made billion-dollar investments in OpenAI and integrated the technology developed by ChatGPT manufacturer into its portfolio of productivity products.

Current Issue