A world too big for one mind
Living as a human, you’ll feel lonely at some point. It just happens.
And in those moments, when I don’t really know who to talk to, I found myself opening an AI and venting to it. At times, what it says really hitting me. Then I thought that, why would AI be useful for personal advice? Maybe, loneliness, not knowing what to do next, feeling a bit unseen. Humans have probably been dealing with the exact same things for thousands of years and people were writing about this long ago. So AI feels really helpful at this scenario.
Then I wish to bring the shared understanding between me and the model across conversations as people usually share the connections. Then I wanna build a memory layer for an AI.
Naturally we’ll start looking at how other people do it. And a few of the tools I came across made me pause. Some of them try to give the AI a forgetting curve. So it slowly forgets things over time, the way a human brain lets memories fade. Some go further and trying to turn AI to feels like human in tone and emotions. Then I wonder about the direction of building my own thing. Should a machine forget the way we do? Should a machine tries to be human? I read somewhere that forgetting is actually a human virtue, not a flaw. We forget so we can move on, get a second chance, not be crushed under everything we’ve ever done.
Around the same time I read a couple of things that reframed the whole question for me.
One is an old idea, the extended mind. Roughly, that your mind might not stop at your skull. The things you think with, a notebook, your phone, can actually become part of how you think, not just things you think about. So the machine isn’t only something I talk to. It can be a part of my mind.
The other is this thing from chess they call the centaur. After Kasparov lost to Deep Blue, instead of being bitter about it, he tried something else. Let a human and a machine play on the same side. In one of those freestyle tournaments, the winners weren’t grandmasters, and weren’t the biggest computers either. They were players who were just really good at knowing when to trust the machine and when to override it. Then it confirms my feeling on the point that the centaur isn’t that the machine is strong. It’s that the human who knows how to steer it is the one who wins.
So here’s what I actually believe, and it’s the reason I lean the way I do.
The world is basically limitless possibility. Too big, too unpredictable. There’s no computing your way through all of it. And humans, somehow, got really good at moving through that mess anyway. Not by calculating everything, but by instinct. By gut feeling. We narrow things down fast, and we’re usually roughly right.
The machine is good at the opposite thing. It can explore a space of possibilities incredibly fast, way faster than I ever could. But the space is so huge that no amount of computing is ever really enough to cover it.
So to me the answer was never a machine that imitates a human. It’s the combination. I use my instinct to point at a small part of the space, the part that’s actually worth looking at. And the machine goes deep into that part, chasing details I’d never have the patience or the speed for.
Which is why I’m building the memory layer more like a mirror, not a replacement for the people I’d rather be talking to. It holds the context, the patterns, the things I keep circling back to, and gives them back to me so I can actually see myself. Reflect on who I’m becoming, and slowly get a bit better at it. It shows me mine, clearer than I could hold it on my own.