Most intellectual labour is proof of work
Haven't you heard? You don't need to write things any longer
I think the idea that LLMs represent meaningful progress in the field of intellectual work purely for their ability to generate textual documents is a category error.
You see, the production of most documents - financial records, contract bids, licenses, documentation, progress reports, interface control documents, etc. - has very little to do with the document itself. Instead, their existence acts as a form of intellectual proof of work, evidence that the author has taken the time and energy to carefully consider their subject matter. The quality of the final document is nothing but a signal that exists solely for the purpose of distinguishing the author’s efforts from background noise.
Why do we need to do this? Because the quality of intellectual work is extremely difficult to measure. A layman can evaluate the craftmanship and care with which a builder constructs a house without much effort: but evaluating intellectual work usually requires the eye of a subject matter expert, and yet decisions about quality must be made by managers who are - by definition - not subject matter experts. So, the document becomes the proxy by which the work of a subordinate expert might be measured.
The age of LLMs has ended the ability for document quality to act as a proxy for intellectual quality. It’s now trivial to kick out a passable document for any given subject, and the document need not attest to any underlying intellectual labour being performed at all. The irony is that by leaning into LLMs so heavily, many organisations have managed to demolish the very signalling system that allowed them to measure their own progress and promote talent.
This is, I believe, why this so-called revolution doesn’t seem to have made a midge’s armpit of difference to the growth, well-being, stability, or capability of the organisations that have leaned into it: because it was never the production of documents that was the true source of value, but the intellectual labour that document-writing requires along the way.