How to Be Irreplaceable

There's a strange irony sitting at the center of Alberto Romero's latest essay for The Algorithmic Bridge: the more precisely you can explain your job, the faster you're teaching a machine to take it. Every process document, every KPI, every neatly systematized workflow becomes training data. You spend years making your expertise legible, and legibility turns out to be the exact thing large language models feed on.

Romero's argument leans hard on a distinction that AI labs themselves are wrestling with. Coding and math have seen genuinely superhuman leaps because those domains have clean feedback loops — you're either right or you're not, and that clarity lets reinforcement learning grind toward mastery. Writing hasn't had its equivalent breakthrough. He cites Erik Hoel's observation that six years after GPT-3, there's still been no 'move 37' moment for prose, and quotes several other writers — Jasmine Sun, Adam Mastroianni, Sam Kriss — all converging on the same point: authorial voice comes from a body that eats, ages, grieves, and gets things wrong in specific, textured ways. AI has no meal to taste, only the cookbook.

The practical upshot, according to Romero, is that the undescribable parts of a person — intuition, taste, the accumulated weirdness of a lived life — aren't leftover fluff. They're the actual moat. Anything you can fully articulate as a procedure is, almost by definition, automatable. The skills companies spent decades training out of workers because they didn't fit neatly into a spreadsheet are precisely the skills that don't transfer to a model, because nobody can hand-hold an algorithm through 'a feeling in your gut' the way you can hand-hold it through a math proof.

What's notable is that this isn't presented as comforting speculation about AI's limits so much as a bet on an open technical question. Nobody inside the top labs actually knows whether the coding-and-math trick generalizes everywhere, or whether describability is a hard ceiling. Romero is staking his advice — the piece promises five principles drawn from history's most 'undescribable' figures — on the second answer holding.