Microsoft Develops MAI-1, a new AI to Rival Google and OpenAI

  Pexels

Microsoft is reportedly developing a new artificial intelligence (AI) model, internally referred to as MAI-1, that aims to compete with similar models from Google and OpenAI, according to The Information. This follows the trend of tech giants investing heavily in AI technology to stay competitive.

New Model Under Development

The new AI model is being overseen by Mustafa Suleyman, the co-founder of Google DeepMind and former CEO of AI startup Inflection. The objective of the model is yet to be determined and will depend on its performance. There are hints that Microsoft might preview the model at its upcoming Build developer conference.

Increased Investment in AI

MAI-1 is expected to be “far larger” than the open source models previously trained by Microsoft, which means it will also be more expensive. This development comes after Microsoft launched a smaller AI model, Phi-3-mini, last month, aiming to attract a wider client base with cost-effective options.

Microsoft has shown its commitment to AI by investing billions in OpenAI and deploying the ChatGPT maker’s technology across its productivity software suite. This move has given Microsoft an early advantage in the generative AI race.

High-Tech Training

The tech giant is reportedly using large clusters of servers equipped with Nvidia’s graphic processing units and vast amounts of data to improve MAI-1. The model is expected to have roughly 500 billion parameters. For comparison, OpenAI’s GPT-4 has one trillion parameters, and the recently launched Phi-3-mini has 3.8 billion parameters.

Suleyman was appointed in March as the head of Microsoft’s newly created consumer AI unit, and several employees from his former company, Inflection, were also hired. The new model does not originate from Inflection, although it might utilize training data from the startup.