By Jeffrey Dastin

SAN FRANCISCO (Reuters) -Microsoft is forming a new team that wants to build artificial intelligence that is vastly more capable than humans in certain domains, starting with medical diagnostics, the executive leading the effort told Reuters.

Called the MAI Superintelligence Team, the project follows similar efforts by Meta Platforms, Safe Superintelligence Inc and others that have begun targeting technical leaps while garnering skepticism for their ability to deliver, absent new breakthroughs.

Microsoft plans to invest “a lot of money” on the project as well, said Mustafa Suleyman, the AI chief in charge.

Meta this year offered $100 million signing bonuses to recruit famous AI talent. Suleyman declined to say if such offers or poaching attempts were on the table. However, he said Microsoft AI would continue to recruit from other top labs while staffing its new team with existing researchers and Karen Simonyan as chief scientist.

Microsoft’s effort comes with a twist. According to Suleyman, the company is not chasing “infinitely capable generalist” AI like some peers. The reason, he said, is he doubts that autonomous, self-improving machines could be controlled, despite research into how humanity might keep AI in check.

He said Microsoft has a vision for “humanist superintelligence,” or technology that could solve defined problems with a real-world benefit.

“Humanism requires us to always ask the question: does this technology serve human interests?” said Suleyman.

AI theorists and developers have long debated whether the technology may lead to imminent danger or poses no harm relative to problems such as machine-learned bias and trustworthiness.

Suleyman said he aims to focus the Microsoft team on specialist models that achieve what he called superhuman performance while posing “virtually no existential risk whatsoever.” He gave as examples AI that solves battery storage or develops molecules, in a nod to AlphaFold, DeepMind’s AI models that can predict protein structures. Suleyman was a DeepMind co-founder.

Suleyman said that for diagnosis, a domain long of interest to the AI field and one that Microsoft has focused on, the company has a “line of sight to medical superintelligence in the next two to three years.”

He said the effort is based on AI that reasons through problems and still would require breakthroughs. But if achieved, he said the AI would “increase our life expectancy and give everybody more healthy years, because we’ll be able to detect preventable diseases much earlier.”

(Reporting by Jeffrey Dastin in San Francisco; Editing by David Gregorio)