So this ad in my feed claims that we're inundated with AI slop because half of the internet is AI-generated and we can't tell which half. Which pretty much means the other half of the internet is human-generated slop?
Iow, with the right tools you can replace ten times as many managers as coders, and you even do the world a favor.
R to @DrDaronAcemoglu: Thanks for the ping @CemFDagdelen!
R to @DrDaronAcemoglu: If you don't realize that knowledge work is a three-step process bc all you ever learned is the middle step, then you can come to the conclusion that everything you know will be automated away. It's just, well, a parochial view. In any case, paper is here. https://drive.google.com/file/d/1XWjBrTsLm-dxoyHXcr_JvgYw23IiPPkg/view?usp=sharing
R to @DrDaronAcemoglu: My favorite Drucker quote applies here, "The first task of knowledge work is to find out what the task is." Problem is that academia has stopped producing knowledge workers in the 1960s in favor of strict formalizers. Herb Simon used to complain about that.
R to @DrDaronAcemoglu: I understand the remonstrations of academics, the Baumol cost disease that kept them gainfully employed is currently being washed away at brutal speed. I produce enterprise grade modeling in hours to days now. If all you can do is that middle part slowly you're in deep trouble.
R to @DrDaronAcemoglu: Once you introduce that starting point, and conjecture that the competitive advantage of machines lies in the "operational middle", the human comparative advantage falls out immediately: the beginning and the end, aka the impetus and the sign-off.
R to @DrDaronAcemoglu: We've been thru a bunch of "computers will kill us all" scares before, and somehow we still made it thru ok. Human effort and cognitive specialization are endogenous and via Schumpeterian recombination we tend to find niches of comparative advantage that keep us employed.
R to @DrDaronAcemoglu: Tbf the total effort (mine plus Gemini's) was about half an hour on the way to a concert and between sets. This wasn't run thru any enterprise grade tools, and I used Gemini bc the other LLMs were busy with more important stuff. But even at this stage it pinpoints the fatal flaw.
I was asked what I think of @DrDaronAcemoglu's new AI paper, so I outsourced the task to @GeminiApp: "Attempting to save human cognition by mandating operational friction is the intellectual equivalent of taxing tractors to preserve the agronomic knowledge of manual plowing."
Crucially, this is not biological exceptionalism. If an artificial system ever becomes conscious, it will be because we engineered the correct intrinsic physical dynamics (the territory), not because we ran a sufficiently complex algorithm (the map). 5/5
The biggest problem with enterprise LLM adoption is that the output resembles craft beer: sometimes stellar, too often just some weird murky concoction, when industry mostly needs dependable ok-ish Bud Light.
I always found "everyone stops looking when the answer matches their priors" the most parsimonious of all world models.
The increasingly common fail mode for "PhD level intelligence" frontier LLMs is "can't be arsed".
Just a PSA that @claudeai destroys whole chat histories. I guess that's the kind of stuff that happens when you're a multibillion $$$ company and you vibecode your whole codebase.
Essential reading from @oliverbeige
#stocks #flows #transformations #operationsresearch
#economics #statistics #compsci #appliedmathematics
https://econpatterns.substack.com/p/stocks-flows-transformations-the
The lesson of the last few years should be that almost everyone is perfectly fine with the state turning sinister, as long as it turns sinister in their preferred direction.
Charles C. Holt, engineer, economist, forecaster, now out on Seemingly Incompatible: Bounding Rationality at Carnegie Tech, 1949-74.
https://carnegietech.substack.com/p/charles-c-holt-engineer-economist #econtwitter
Vendors can impose terms of use on their products as part of a sales contract, prospective buyers can seek alternative vendors if they disagree with the terms. Nothing to see here except everyone going completely overboard.
If you feel compelled to split your workforce in half "because AI", you should consider splitting them into "new thing" and "old thing" instead of "unemployed" and "still employed", especially if you have access to risk capital and you're not running out of ideas @jack & @blocks.
R to @oliverbeige: COBOL was never the bottleneck. Despite its unhip reputation, it's gotten the job done for 60+ years, and there's no reason to replace it bc of any problems with COBOL itself. The only reason was that there weren't any coders left. And that's solved now.
https://x.com/i/status/1906054100669612219
Controversial take: that COBOL vibecoding thingy is actually good news for IBM.
I expect a hard reversion into professional knowledge, bc that's what the machines can't automate.
The last 40 years were essentially a long run towards replacing professional knowledge with a cheaper simulation thereof, with mixed results on the micro level and disastrous results on the macro level.
R to @oliverbeige: Norbert Wiener, The Human Use of Human Beings, 1950. Two years after Cybernetics, things were already going down the drain.
https://archive.org/details/humanuseorhumanb0000norb/page/188/mode/2up
Norbert Wiener, merchant of doom.
Parallel Paths to Cognition: The Carnegie-UCSD Connection in Artificial Intelligence is out now on Seemingly Incompatible.
https://carnegietech.substack.com/p/parallel-paths-to-cognition-the-carnegie #econtwitter #machinelearning
R to @oliverbeige: And that's really all you need to know.
It's because LLMs use gradient descent to traverse jagged search spaces.
Almost every criticism of AIs supposed inadequacy is a reflection of the critic’s own inability to prompt AI effectively.
If you hammer away at a piano for a week and get upset it cant produce Chopin. Thats on you.
Shattering the knowledge frontier.
The whole "soon we're all gonna be unemployed and unemployable" doomsaying also played out at the dawn of the personal computer age, but luckily nature provided infinite work for everyone in the form of compliance documentation and exploding administrative overhead.
Herb Simon and Al Newell's "General Problem Solver", according to Grok.
New blog post:
Autodiff through function types: Categorical semantics the ultimate backpropagator, by @_julesh_
https://cybercat.institute/2026/02/20/categorical-semantics-ultimate-backpropagator/
Gemini is getting tired of reasoning...
The Library of Babel: Combinatorial Optimization at Carnegie is now out on Seemingly Incompatible.
https://carnegietech.substack.com/p/the-library-of-babel-combinatorial #econtwitter
"The words or the language do not seem to play any role in my mechanism of thought. The psychical entities which seem to serve as elements in thought are certain signs and more or less clear images which can be voluntarily reproduced and combined." — Einstein to Hadamard, 1945.
The reason why the Ringbahn works so well as a synecdoche for the attention economy is that Berlin's tariff area A is all attention economy, no information economy. Which is also the reason why Berlin comes across like a circus sideshow.
There are two alternatives:
- Handwritten exams with zero technology aid in a tightly controlled environment.
- You can use whatever you want, but the task is tailored so that it cannot be solved by an LLM alone, and a human alone would need months or years to solve it.
"A Science of the Artificial: Herbert Simon on Design" is now out on Seemingly Incompatible.
https://open.substack.com/pub/carnegietech/p/a-science-of-the-artificial-herbert