Microsoft is in the process of developing a new, energy-intensive generative AI model named MAI-1, as reported by The Information. The model is being spearheaded by Mustafa Suleyman, a former Google AI executive and recent addition to Microsoft from AI firm Inflection, which Microsoft acquired earlier this year. This new AI venture is distinct from Inflection’s existing Pi models.
To facilitate the training of MAI-1, Microsoft is reportedly amassing a significant number of servers equipped with Nvidia graphics cards. The size and complexity of MAI-1 are expected to exceed Microsoft’s previous AI endeavors, requiring substantial electricity which raises environmental concerns due to the high energy consumption.
In a related blog post, Microsoft CTO Kevin Scott emphasized the company’s ongoing commitment to developing advanced AI technologies, mentioning other AI models like Turing alongside MAI. Scott’s message seems to downplay the novelty of this development, reiterating that building powerful supercomputers for AI training is standard practice for Microsoft.
The tech giant continues to aggressively expand its AI portfolio, recently integrating new features such as a dedicated Copilot AI key in Windows 11 and priority access to OpenAI’s GPT-4 Turbo for its 365 subscribers. These initiatives are part of Microsoft’s broader strategy to maintain a competitive edge over other tech giants like Amazon, Apple, and Google.
However, Microsoft’s rapid advancement in AI has not been without controversy. Recent internal and public criticisms have focused on issues ranging from the generation of inappropriate content by its AI tools to the methods used in its toxicity testing.
