
Mustafa Suleyman announced that his Microsoft AI division has created a team pushing towards Humanist Superintelligence (HSI). The goal? To build practical technology that is designed specifically to serve humanity.
“We think of Humanist Superintelligence (HSI) as systems that are problem-oriented and tend towards the domain specific,” Mr. Suleyman writes in the announcement post. “Not an unbounded and unlimited entity with high degrees of autonomy – but AI that is carefully calibrated, contextualized, within limits. We want to both explore and prioritize how the most advanced forms of AI can keep humanity in control while at the same time accelerating our path towards tackling our most pressing global challenges.”
Suleyman positions HSI as a practical alternative to Artificial General Intelligence (AGI), the stated goal of OpenAI, and a rather nebulous concept in which AI will somehow “match human performance at all tasks.” Indeed, he outright rejects the race to AGI, as he does the “binaries of boom and doom” that define the debates around AGI. HSI is about AI servicing humanity, not potentially replacing it.
To get to HSI, he has created and will lead a Microsoft AI Superintelligence Team “to be the world’s best place to research and build AI, bar none.” It will solve “real concrete problems,” and do so in a “way that it remains grounded and controllable.” This is not “an ill-defined and ethereal superintelligence,” he says, but rather “practical technology explicitly designed only to serve humanity.”
Superintelligence can continuously improve itself, he notes, so it will need to be contained and aligned with human progress not just once, but constantly, in perpetuity. This is an “urgent challenge,” Sulleyman adds, and so this team will prioritize the interests of humans over any research or development agenda.
In many ways, then, this reads as a refutation of OpenAI and its goals, which is interesting given the partnership between that company and his employer. But Microsoft AI has been arguably working to reduce and then eliminate Microsoft’s reliance on OpenAI, despite public protestations to the contrary. And this new goal certainly aligns with that need.