Microsoft is officially joining the “we’re totally not building Skynet” club. Last month, Mustafa Suleyman, the company’s Head of AI (and part-time philosopher of machine ethics), announced the creation of the MAI Superintelligence Team, a group that will conduct “advanced artificial intelligence research.”
The name sounds like something out of a futuristic movie, and the goal sounds like something out of a PR manual: solve real problems “in a grounded and controllable way.” In other words, don’t panic, the robots are here to help, not to take over.
The Man Behind the Machines
Suleyman, who co-founded DeepMind (the AI lab Google bought in 2014) and later started Inflection AI, is now leading this new venture under Microsoft’s AI division, home to Bing, Copilot, and every overly polite chatbot that insists it “understands your frustration.”
In his official blog post, Suleyman assured everyone that this isn’t an attempt to build a digital deity. He wrote, “We are not building an ill-defined and ethereal superintelligence, we are building a practical technology explicitly designed only to serve humanity.” You know it’s serious when the CEO has to clarify, “Don’t worry, it’s not a robot overlord.”
The Great AI Arms Race, Round Two
Microsoft’s announcement comes just months after Meta, yes, the same company still trying to make the metaverse happen, launched Meta Superintelligence Labs, its own attempt at making machines smarter than humans.
“Superintelligence” is tech-speak for AI that’s supposed to surpass human brainpower, a comforting idea for anyone who’s ever yelled at a virtual assistant for misunderstanding “set a timer.”
And because no major tech company likes to feel left out, Microsoft had to jump in. With OpenAI, Anthropic, Google DeepMind, and Meta all competing to see who can build the first machine that knows it’s smarter than us, this move feels inevitable, like a group project where everyone wants credit for doing the most work.
Microsoft’s Web of AI Relationships (It’s Complicated)
Here’s where it gets interesting. Microsoft’s relationship with OpenAI is less a partnership and more a high-tech situationship. Microsoft uses OpenAI’s models to power Bing and Copilot, while OpenAI runs its workloads on Azure. Oh, and Microsoft owns a casual $135 billion equity stake in the company. They’re not quite married, but they definitely share assets.
Still, Microsoft has been quietly hedging its bets. After buying parts of Inflection, it began using AI models from Google and Anthropic too, because commitment is hard when innovation moves this fast.
What the New Team Will Actually Do
According to Suleyman, the MAI Superintelligence Team will focus on building “useful companions” that can assist with education, medicine, and renewable energy. He claims the group aims for “expert-level performance in diagnostics and highly capable planning and prediction in operational clinical settings.”
So, think of it as Microsoft’s attempt to make an AI that can help you with your homework, your health, and your electricity bill, all before reminding you to update Windows.
Investors Want Superprofits, Not Superbrains
Of course, investors aren’t interested in philosophical debates about serving humanity. They want to know when all this “superintelligence” will start paying off.
Suleyman tried to calm the growing skepticism, saying the company is “not building a superintelligence at any cost, with no limits.” It’s a reassuring statement that still manages to sound like someone promising their experimental rocket is “mostly safe.”
The Bottom Line
So, Microsoft’s building a “practical” superintelligence. Meta’s building one too. OpenAI’s redefining its moral boundaries. And Google’s somewhere in the background, quietly muttering, “We were doing this first.” It’s starting to feel less like a tech race and more like a high-budget drama where everyone insists their creation will “serve humanity,” right before it learns to code itself.
Suleyman promises Microsoft’s new project won’t spiral into something uncontrollable. Let’s hope he’s right. Because if history has taught us anything, it’s that the sentence “We can control it” never ages well
