After finishing the Week 17 newsletter, I sat staring at nothing for a moment. Not because the work was done, and not exactly from fatigue — though I was tired. It was a stranger kind of exhaustion: like finishing a long run and finding the road ahead has doubled, branched, and sprouted new signs, while you’re still standing there with the same breath, the same back, the same eyes.
AI is advancing too quickly. Quickly in a way that is no longer news, but weather. You open your eyes to a new model, a new capability, a new demo, a new anxiety. At some point you’re no longer reading about AI — you’re being read by it, measured against your slowness, your confusion, your struggle to hold a human pace against something accelerating with excessive politeness.
In that moment of staring, I thought: maybe AI doesn’t make people better. It only amplifies them.
Call it a Human Amplifier — a slightly Western-sounding term for something actually quite mundane: a loudspeaker placed in front of whatever already exists inside a person. No salvation, no enlightenment — it doesn’t automatically make anyone more decent or wiser. It only makes what’s already there louder, sharper, smoother, and sometimes more dangerous.
The people who seem to get the most from it are the ones who know how to doubt what it gives them. They use it as a mirror to find the places they hadn’t thought to look, and they feel faintly awkward before a sentence that came out too polished. (The smoother sentences in my newsletter are from an AI assistant. I supplied the ideas.) For people already in the habit of questioning their own assumptions, a few exchanges with AI can be humbling: place what you know beside a system trained on far more text than any person will read in a lifetime, and your knowledge starts to look small — small, but not necessarily bad. Small enough to keep learning.
The ambitious are different, and I think this is where it gets genuinely tricky. AI can give someone the feeling of having improved very quickly — writing better, speaking more coherently, producing sharper critiques. I know this feeling. There are weeks in the newsletter where I’ve caught myself accepting a well-constructed paragraph and moving on, quietly setting aside the question of whether the thought behind it was actually mine. That’s the shadow on the wall mistaken for the real body: credit claimed, responsibility for thought quietly set aside.
And then there is the version that worries me most in other people, because I suspect I’m not immune to it either. Someone left a critique on one of my articles recently, and when I reread it carefully, it read like unedited AI output — the kind of response that is structurally coherent, covers the expected objections, and says almost nothing the person chose to mean. I stopped replying. There’s not much point in debating someone else’s model. But the thing that stayed with me wasn’t the rudeness of it; it was how easy it is to produce the surface texture of a considered opinion without the cost of actually forming one. Foolishness has always existed. AI just gives it citations and paragraph structure.
All of this makes me uneasy — because after this particular newsletter, after hours of reading, filtering, writing, editing, asking AI for an idea and then discarding it, I’m probably being amplified in ways that aren’t entirely flattering either. The self-doubt arrives here, at the end, which is a little too tidy. I should say it plainly: I don’t know which parts of me are growing and which parts are just getting louder. The impatience is real. The desire to keep up, the fear of being left behind, the urge to publish fast enough to match something I haven’t fully understood yet — those are in the newsletter too, somewhere, even when I can’t point to the sentence.
AI doesn’t replace growth so much as amplify whatever already exists in us; and maybe the genuinely unsettling thing isn’t that it will surpass humans — it’s that it seems to help each person become themselves faster, even when that self has never been closely examined, never been practiced, never been made decent.
I wrote this in a state of uncertainty. There was AI in the drafting. There was me in the hesitation. I still don’t know which part of me this paragraph is amplifying.