Microsoft quietly builds a ‘superhuman AI’ to rival OpenAI, Meta

Microsoft has announced the launch of a new division dedicated to building AI systems that can outperform humans in specialised areas, starting with medical diagnosis. 

The initiative, called the "MAI Superintelligence Team," marks one of the company’s most ambitious research pushes yet, led by Microsoft AI chief Mustafa Suleyman.

According to Reuters, Suleyman stated the company plans to invest “a lot of money” into the effort, signalling a deep commitment to developing superhuman AI models that can solve real-world problems. 

Unlike rivals rushing toward artificial general intelligence (AGI), Suleyman stressed that Microsoft is not pursuing an “infinitely capable generalist machine.” 

Instead, the focus is on what he calls “humanist superintelligence,” powerful AI systems designed with strict boundaries and intended to advance human health, science, and engineering.

The new team includes top researchers from Microsoft and will be led scientifically by renowned AI expert Karen Simonyan. 

Suleyman did not mention whether Microsoft will offer massive signing bonuses like Meta’s reported $100 million incentives but hinted that recruitment from top labs would continue.

He said the company sees a clear pathway to “medical superintelligence” within the next two to three years, with AI capable of outperforming human doctors in certain diagnostic tasks.