SAN FRANCISCO, Nov 6 (Reuters): Microsoft has announced the creation of a new artificial intelligence research unit aimed at developing “superintelligent” systems capable of outperforming humans in specialized fields, beginning with medical diagnostics.
Aiming for human-level breakthroughs
The initiative, known as the MAI Superintelligence Team, will focus on building advanced AI models designed to deliver measurable real-world benefits rather than pursuing fully autonomous general intelligence. The project is being led by Mustafa Suleyman, Microsoft’s head of AI, who told Reuters that the company plans to invest “a lot of money” into the effort.
The announcement follows similar pursuits by Meta Platforms, Safe Superintelligence Inc, and others racing to push the limits of AI performance. However, many experts remain skeptical about whether true superintelligence can be achieved without fundamental scientific breakthroughs.
Humanist superintelligence
Suleyman, who co-founded Google DeepMind, said Microsoft’s vision differs from competitors that are chasing all-purpose AI systems. Instead, the company is focusing on what he calls “humanist superintelligence” — technology that advances human welfare without introducing uncontrollable risks.
“Humanism requires us to always ask the question: does this technology serve human interests?” he said.
He explained that Microsoft’s goal is to build specialized AI that can achieve superhuman performance in certain problem domains while posing “virtually no existential risk.” Examples include AI that accelerates battery innovation or discovers new medical compounds — areas that can drive societal and scientific progress.
Medical superintelligence within years
Medical diagnosis will be the team’s first target. Suleyman said Microsoft has “a line of sight to medical superintelligence in the next two to three years,” suggesting the company believes AI could soon surpass human capabilities in identifying diseases.
The envisioned systems would use reasoning-based models to interpret symptoms, predict conditions, and flag early signs of illness. Suleyman said that if successful, such AI could “increase life expectancy and give everyone more healthy years by detecting preventable diseases much earlier.”
A new direction in AI research
Microsoft’s superintelligence initiative reflects a growing debate among AI theorists about how far—and how fast—machines should advance. While some warn of existential risks from unchecked AI, others argue that focused, human-guided systems can transform industries safely.
Suleyman said the company will recruit top researchers from leading labs and has appointed Karen Simonyan as chief scientist of the new division. The group’s early projects will emphasize safety, transparency, and real-world applications before expanding to broader challenges in science and technology.
This news was originally published by Reuters.