February 25, 2026
Ants, AIs, and academic egos
I don't know how you get here from "predict the next word."
Prof says AI gave best feedback ever — awe, ant metaphors, and “finance is gobbledygook”
TLDR: An economist says an AI editing tool gave him stunning, peer‑level feedback on his inflation paper and told him to make claims testable and clearer. Commenters split between marveling at “emergent” smarts, dunking on finance as fluff, and warning that AI‑polished writing could supercharge persuasion over substance.
An economist fed his inflation booklet to a new AI editing tool, refine, and came back starry‑eyed: the feedback was sharper than most peer reviews he’s seen. He insists he didn’t use AI to write the piece—just to critique it—and the tool told him to make his claims testable and clarify the economic model drama. Cue the comments section, which went full popcorn mode.
On Team Awe, users called it “emergent behavior”, arguing that while large language models (LLMs) technically “predict the next word,” we still don’t know how complex smarts pop out—think ant colonies making cities. One commenter pointed to 3Blue1Brown’s explainer and said, essentially, the machines got big and weird and now they show new tricks.
On Team Meh, a savage line landed: finance papers are “gobbledygook and extensive filler,” so maybe the bar’s low. Another user’s “neo‑Sophistry” alarm went off, worried that slick AI feedback will make arguments sound irresistibly polished, even when the logic’s shaky. Meanwhile, a calmer voice believed the author didn’t use AI to write, just to edit, and that’s fine.
Translation for non‑wonks: an AI editor told a professor to pin his inflation story to real dates and tidy his model talk. The real show is the thread: miracle ants vs. word parrots, with a side of “is AI making us too smooth to trust?”
Key Points
- •The author tested the AI tool “refine” on a draft booklet about inflation and received structured, high-quality feedback.
- •The tool warned the manuscript’s “fiscal news” narrative risks circularity and urged using dated, external observables tied to inflation expectations and long-bond prices.
- •It recommended clarifying the distinction between FTPL and New Keynesian models by focusing on fiscal regime closure and the passive fiscal assumption in NK.
- •The author acknowledged limitations in measuring present value of surpluses and aims for plausibility, planning to tighten evidence and framing.
- •The author noted tension between observational equivalence and model identification, proposing use of non-time-series information about central bank behavior to assess plausible regimes.