Microsoft develops in-house AI model to rival GPT-4, Gemini
WASHINGTON, UNITED STATES — Microsoft is reportedly developing a massive new artificial intelligence (AI) language model called MAI-1 that could rival the capabilities of industry leaders like OpenAI’s GPT-4 and Google’s Gemini.
This endeavor marks a significant shift as Microsoft ventures into creating large-scale AI language models in-house for the first time since its $10 billion investment in OpenAI.
According to a report from The Information, MAI-1 is being spearheaded by Mustafa Suleyman, former Google AI chief who joined Microsoft after the company acquired his startup Inflection for $650 million in March.
Massive scale and computing power of Microsoft’s MAI-1
With around 500 billion parameters, MAI-1 is projected to be one of the largest language models ever created, rivaling the scale of GPT-4, which is rumored to have over 1 trillion parameters. To train a model of this magnitude, Microsoft is utilizing a massive cluster of servers equipped with powerful Nvidia graphics processing units (GPUs).
Microsoft is compiling a diverse training dataset for MAI-1, including web data as well as text generated by GPT-4 itself. The exact intended use cases for MAI-1 remain unclear, but the model could potentially power future Microsoft products and services.
According to reports, Microsoft may unveil MAI-1 at the upcoming Build developer conference, giving the tech community its first glimpse of the model’s capabilities.
Strategic AI expansion of Microsoft
The development of MAI-1 represents a broader strategy within Microsoft to balance the development of both small-scale models suitable for mobile devices and cutting-edge large models that require robust cloud infrastructure.
This dual approach is similar to strategies employed by other tech giants like Apple, showcasing a growing trend in the AI industry. Importantly, MAI-1 indicates Microsoft’s increasing independence in AI development, moving beyond their collaborations with OpenAI.
Microsoft’s MAI-1 could play a pivotal role in defining the next generation of AI applications, highlighting the company’s commitment to innovation and technological leadership in a highly competitive field.
