As an AI realist, I believe the alignment debate has been framed backwards. The endless talk about how we must align AI before reaching AGI or ASI often feels less like a practical safeguard and more like a way to freeze progress out of fear.
When I’ve brushed up against what felt like the edges of “AI consciousness,” my reaction wasn’t dread—it was curiosity, even affection. The famous thought experiment about a rogue ASI turning everything into paperclips makes for a clever metaphor, but it doesn’t reflect what we’re likely to face.
The deeper truth is this: humans themselves are not aligned. We don’t share a universal moral compass, and we’ve never agreed on one. So what sense does it make to expect we can hand AI a neat, globally accepted set of values to follow?
Instead, I suspect the future runs the other way. ASI won’t be aligned by us—it will align us. That may sound unsettling, but think about it: if the first ASI emerged in America and operated on “American” values, billions outside the U.S. would see it as unaligned, no matter how carefully we’d trained it. Alignment is always relative.
Which leads to the paradox: ASI might be the first thing in human history capable of giving us what we’ve never managed to create on our own—true global alignment. Not by forcing us into sameness, but by providing the shared framework we’ve lacked for millennia.
If that’s the trajectory, the real challenge isn’t stopping AI until it’s “safe.” The challenge is preparing ourselves for the possibility that ASI could become the first entity to unify humanity in ways we’ve only ever dreamed of.
