Microsoft AI chief Mustafa Suleyman shifts focus to superintelligence

In short: Microsoft AI CEO Mustafa Suleyman says he has been shifting his role toward building “superintelligence” and reducing Microsoft’s reliance on OpenAI.

What happened

Mustafa Suleyman, the CEO of Microsoft AI, told The Verge he has been preparing for a strategic shift in his job for up to nine months. The change became public after a restructuring inside Microsoft in mid-March.

Suleyman was appointed Executive Vice President and CEO of a newly created consumer AI unit in March 2024. He says a central goal is to build Microsoft’s own “foundation models” (large AI systems that other tools can be built on), so Microsoft does not depend as much on OpenAI.

One key step was renegotiating Microsoft’s long-term relationship with OpenAI. That deal extends Microsoft’s license to use OpenAI’s intellectual property, meaning its protected technology and know-how, through 2032.

Suleyman also described a future where making a new AI model could become much easier. He compared it to creating a podcast or writing a blog, with organizations able to make custom AI systems for their specific needs.

He said winning in AI may depend less on who has the “smartest” model, and more on who can run AI quickly and cheaply. This is often called “inference,” which is when an AI gives an answer after it has been trained (like using a calculator, not building it).

Suleyman also predicted AI could reach human-level performance on many professional tasks in 12 to 18 months. He specifically pointed to computer-based office work in areas like accounting, law, marketing, and project management.

Why it matters

For regular people, this points to two practical changes. First, AI features in everyday products like work software could get faster and cheaper. Second, some office tasks may be automated sooner than many companies and workers expect, which could reshape how white-collar jobs are done.

Source: The Verge AI