Here’s a working example of the LLM-as-research-partner loop most people only talk about. Wilson, an independent blogger with no formal economics background, spent months developing a theory of wages with Anthropic’s Claude as a working partner — the model (named “Fable 5” in the essay’s now-removed acknowledgements) fetched data, caught wrong assumptions, and surfaced papers showing the opposite. The result is now being formalized with a co-author at the Stockholm School of Economics, with a paper on SSRN.
The economics: standard theory doesn’t actually know how aggregate wages are set — wage models are estimates with free parameters. Wilson’s move was to take Acemoglu and Restrepo’s task-based framework (where the wage is set at the marginal task, w = c·γ(x*), the point where humans and machines are about equally good) and add the classical economists’ input-output recursion on top. Machines cost machines, labor, and land to build, so the machine’s rental price contains itself: recurse it out and the wage resolves to w = γ(x*)·br / (1 − a − λγ(x*)) — set by technology and access to physically scarce things (land, oil, ore), scaled by how efficiently machines make machines. Nothing in the machinery is new — Leontief and Sraffa’s recursion is from 1936, Ricardo’s rent from 1817 — the claim is that smushing them together pins the wage with no free parameters and “fits history like a glove.”
Two predictions follow. First, housing costs rising relative to everything else shouldn’t surprise anyone: the model says the human edge γ(x*) started falling around the 1970s and really fell once the internet took off, and a falling γ(x*) pushes wages down while rents keep climbing. Second, AI might lower γ(x*) much further and more generally — cognitive automation is exactly the kind of technology that erodes the marginal human task. The policy answer then falls straight out of the math, and it’s also nothing new: Henry George, 1879 — tax the scarce things (land above all) and fund consumption from it. Wilson would add a sovereign wealth fund on top, because owning equity captures other scarcity (network effects) and works across borders in a way land taxes can’t — Norway already runs the playbook.
The uncomfortable part is the last paragraph, reproduced from the paper’s removed acknowledgements: the prose you just read was written by Claude itself, in its own voice, which the author deliberately did not overwrite — Richard Sutton’s bitter lesson would advise as much. An LLM that helped develop an economic theory predicting its own depressing effect on wages, and then wrote the essay about it. Whether Wilson stands on the shoulders of giants or of “something new entirely” is left as the essay’s open question — and it’s the question that makes this worth reading. Follow-up posts on the blog (“There’s a bug in the Economy, and Fable found it,” “Why do we assume everyone should be working?”) keep the thread going, and there’s an HN discussion on the piece.