Microsoft
Microsoft promises its future AI superintelligence won’t eat us
Microsoft plans to keep humans “at the top of the food chain.” (for now)
Just a heads up, if you buy something through our links, we may get a small share of the sale. It’s one of the ways we keep the lights on here. Click here for more.
Microsoft would like everyone to relax: yes, it’s working on superintelligent AI, but no, it’s not building the thing that eventually decides humans are inefficient meat-clogs in the progress pipeline.
In a new blog post, Microsoft AI chief Mustafa Suleyman, formerly of DeepMind and Inflection, announced the company is forming a team dedicated to what he calls “humanist superintelligence.”
AI that’s powerful, helpful, and very much not in charge.
Suleyman insists this future superintelligence will not be the sci-fi nightmare version roaming free with infinite autonomy, but something “carefully calibrated” and “within limits,” like an all-knowing assistant that still has to raise its digital hand before speaking.
It will exist only to “serve humanity,” not replace it, overthrow it, or even sass it (though no promises).
This comes at an interesting time. Microsoft has just secured the legal freedom to build its own AGI (artificial general intelligence), even using OpenAI’s technology if it wants.
So while Suleyman says “we reject the race narrative,” Microsoft is now fully allowed to win that race, solo or with new partners.
Somewhere in San Francisco, an OpenAI lawyer just got heartburn.
The plan for Microsoft’s human-friendly superintelligence has three big use cases:
- Personal AI companions: think Clippy, if Clippy could manage your life instead of just popping up uninvited.
- Healthcare support: AI doctors, but hopefully the kind that don’t tell everyone to hydrate and walk it off.
- Scientific breakthroughs: ideally in areas like clean energy, not inventing new ways for printers to jam.
“Humans matter more than AI,” Suleyman writes, in what may become the most quoted sentence in future robot history textbooks.
The goal is an AI that is “subordinate, controllable,” and definitely not the kind that opens “Pandora’s Box.”
So yes, Microsoft is building superintelligence, but don’t worry, it’s nice superintelligence.
The kind that keeps humans “at the top of the food chain.” For now.
