Microsoft’s new “super intelligence” game plan revolves around business

Mustafa Suleiman has been preparing for his new job description for a long time. Suleiman was the first CEO of Microsoft’s AI division, but after the company underwent a broad restructuring in mid-March, he stepped back from some duties and shifted focus to chasing superintelligence. Although the news only broke last month, he says EdgeHe’s been preparing for the move for up to nine months — and although the renegotiation of Microsoft’s contract with OpenAI is the thing that “officially unleashed (Microsoft’s) ability to pursue superintelligence,” he was planning even before the ink was dry.

“This was a long-term plan,” he said, adding that achieving superintelligence was “purely my focus.”

Superintelligence – along with AGI, or artificial general intelligence – It has a vague and shifting definition in the AI ​​industry. For Suleiman, it’s all about business and productivity. “Superintelligence is about: Are these models capable of delivering product value to the millions of companies that rely on us to deliver world-class language models?” Soliman said. “That’s really our focus. We want to serve developers, enterprises and a lot of consumers.” AI companies face increasing pressure to generate more revenue, and Microsoft’s plans reflect a new strategy In OpenAI also.

Microsoft’s reorganization brought together enterprise and consumer teams under the Copilot AI banner. While Solomon will still work on big-picture strategy, Jacob Andrew, who was previously corporate vice president of product and growth at Microsoft AI, will become executive vice president. Leading Engineering, growth, product, and design initiatives for newly integrated teams. This shift left room for Solomon to devote his time to pursuing superintelligence and developing new AI models for Microsoft at a time when competition among leading AI companies — and the pressure to attract new consumers and enterprise customers — is more intense than ever.

On Thursday, Microsoft debuted a new copy model that it hopes will do just that — and since it’s “half the GPU cost of other modern models,” according to Solomon, it’s a “huge cost saving” for Microsoft.

The company describes MAI-Transcribe-1 as “pushing the boundaries of speech recognition” with its ability to transcribe meetings, caption videos, and analyze call center exchanges in 25 languages. Microsoft’s blog posts announcing the model say it was designed for “challenging” recording conditions including background noise, low-quality audio, and overlapping speech, and was trained on a mix of “human-curated” transcripts and machine-written transcripts. The source recordings are a combination of controlled sound booth data and contractors tasked with recording themselves amidst background noise, from crowded streets to children running, as well as “massive amounts of data from the open web,” Soliman said.

Along with the existing audio and video generation models MAI-Voice-1 and MAI-Image-2, the new transcription model is now available on Microsoft Foundry and as part of the new Microsoft AI Playground. This is the first time these models have been “widely available for commercial use,” according to Microsoft. MAI-Transcribe-1 can handle audio files in MP3, WAV and FLAC formats.

Sulaiman attributes the new model’s performance in tests to a small team of 10 people. The modeling team is “freed from any bureaucracy,” he says, as they have a perimeter team responsible for managing vendors, finding data to download, and more. Microsoft has used a similar strategy to generate sound and images, and other companies have taken similar steps – such as Meta, Amazon, and Google. Workout with To level out their organizations, Anthropic said it is also experimenting with giving small teams of a few developers free access to certain levels of computing to see what they can achieve.

The new copy model is part of Solomon’s goal to deliver “human-centered” AI (a variation of Microsoft’s model) Artificial Intelligence’s favorite buzzword“superhuman intelligence”) which is useful to the average person. “Everyone will have an AI assistant in their pocket that is truly world-class, accountable to them, accountable to them, aligned with their interests, and working on their behalf,” he said.