AI Won't Level the Starting Line. It Widens It.

Over the past few months, headlines like "ordinary people turning their lives around with AI" have been everywhere. The same question keeps showing up in comment sections: I don't have the background, I'm behind — if I just learn to use AI well, can I catch up with people who were already ahead of me?

Buried in that question is a specific hope: leapfrogging. Skipping the years it would normally take to close a gap, jumping straight to roughly the same starting point as everyone else, just by picking up a new tool. That hope is good business — a flood of AI side-hustle courses and case-study bundles are selling exactly this promise: AI is a fair new starting line, everyone gets on the same footing, and it even hints that people stuck at the bottom might get a head start from this wave of tooling.

I'm currently running a few things in parallel — an agent pipeline that writes and drafts my newsletter automatically, an automated system that sizes positions for a quant strategy, and a handful of smaller tools I built for myself. Watching AI try to do the work for me every day, I've come to think the question "can AI help me leapfrog" is backwards.

Hand the same prompt to two different people and the outputs can end up worlds apart. One person can spot at a glance which line in the AI's plan is filler, which number is wrong, which piece of advice only applied three months ago. The other can't tell, and just copies it as-is. That gap isn't created by AI — AI just takes the judgment that was already sitting, invisible, in each person's head, and puts it out in the open. Once it's out in the open, who's stronger and who isn't becomes obvious immediately.

My own newsletter-writing system is a working example of this. AI handles execution: picking the topic, checking for repeats, laying out images, formatting, pushing the draft. It does all of that fast and reliably. But which topic lane to use, what counts as a real core idea worth writing, whether today's angle overlaps with something I wrote last week and needs to change — those are judgment calls I have to make and turn into explicit rules first. AI just follows them. The moment that layer of judgment isn't nailed down in advance, whatever AI writes collapses into the same template: on the night of August 6, 2026, I laid out my last seventeen posts side by side and read through them one by one. By the third piece, the back of my neck went cold — the same binary-contrast title structure, welded in place; every section ending in the same bolded punchline. Read three posts and you could predict the fourth. The blame doesn't belong on the tool. It's on me, for not having thought through what actually counts as a good post before I let the pipeline run.

The flip side is worse. If someone genuinely lacks judgment in a given area, AI won't fill that gap for them — it will just make their mistakes look more polished: someone who doesn't understand system architecture gets code from AI that runs fine, then crashes at the ten-thousandth record, and they can't tell which layer broke; someone who doesn't understand risk gets a return curve from AI that looks smooth, with nobody telling them how much the worst month could cost; someone who doesn't understand their readers gets a hollow sentence from AI dressed up to sound fluent — reads like an opinion, and there's nothing there once you unpack it. I wrote about a related failure mode before: when every external source an AI agent tried to fetch failed, it didn't say it didn't know — it stitched together a confident-sounding answer, complete with version numbers and dates, from stale knowledge. Same underlying mechanism: AI doesn't hedge just because you don't understand something, it follows whatever direction you point it in, and says it with more certainty.

That's why the same tool, in different hands, produces results that keep pulling further apart. For someone who already has a plan, AI maxes out their execution speed; for someone who doesn't, AI maxes out the confidence behind their mistakes — both are accelerating, just in opposite directions.

Next time you're about to lean on AI to get something done, ask yourself one thing first: if AI weren't in the picture, could I explain how to judge right from wrong on this myself? If you can't, what AI hands you is just a faster, better-looking mistake — not a shot at catching up.