Mastodawn

I don't think I have ever read a more concise and yet thorough explanation of my own observations, conclusions, and fears as a 20+ year veteran technology professional and 15 year SRE veteran.

We are actively tearing down the hard fought wins for ourselves and our customers in the chase for the MBAs and business idiots and the institutional knowledge and processes that solved huge problems are the victims of this AI circlejerk.

https://leehanchung.github.io/blogs/2026/04/05/the-ai-great-leap-forward/

The AI Great Leap Forward

In 1958, Mao ordered every village to produce steel. The steel was useless. The crops rotted. Today's top-down AI mandates are producing the same pattern: ba...

Han, Not Solo

@robert I have casually made this same observation but holy shit this is incredibly detailed and thorough

Addison 17h ago

@glyph @robert I have been subsequently informed that the article is likely produced in part or in whole using LLMs: https://lobste.rs/c/2ioo8a

I don't have the means to detect this, and I'm deferring trust to others who do. I'm so sick of this.

The AI Great Leap Forward

5 comments

Lobsters

Robert A. Hill 17h ago

@addison @glyph I mean the guy is an ML/AI researcher and engineer. I don't think it actually hurts his point in the post. I am not even against all usage of these tools, especially personal usage or whatever

Addison 17h ago

@robert @glyph I am. And it makes me unable to engage with this writing, because I cannot truly determine who's opinion I'm reading: a person's, or the emulation of a person's.

Robert A. Hill 16h ago

@addison @glyph Ok, have a good one then.

@addison @robert I have been tricked by AI writing before and this case doesn’t look nearly as clear-cut. it could of course still be (and the author doesn’t give me the highest confidence) but re-reading it is missing a lot of the really obvious tells. I am still willing to give it the benefit of the doubt; this could easily just be a few stylistic tics that someone exposed to a lot of LLM-extruded text might adopt

@addison @robert throwing snippets of this at a bunch of AI detectors seems to show a pretty high likelihood of human authorship. Not that that’s totally dispositive either, but more evidence for at least the benefit of the doubt

Jonathan 1d ago

@robert This post reminds me of the Reddit post where the person realized that the AI had been hallucinating sales numbers which they had been feeding to the board for months and the fellow felt like he was going to throw up.

congusbongusgames 1d ago

@robert small irony that all the images in the blog post are AI generated

Eniko Fox 21h ago

@congusbongus @robert not really, if you go to his about page it says the guy is literally an LLM salesman

@robert This is every company now, tech or otherwise.

> They’ve never computed a baseline.

“How will you know it works?”
“Oh, I’ll know!”

/me screams into the abyss

Although having read some other blog posts there, I wanna scream even more. Especially the one on determinism. Broken clock and all that.

@robert
A great AI journalism article.

And I see it is deeply appreciated by many on the Fedi.
Han Lee is doing good AI journalism and I'm sure we'd welcome him and engage with his AI journalism posts fulsomely.

Sadly there are some AI journalists who feel unwanted, "chased off" even, on the Fedi.
Might that be because what they call AI Journalism is too often indistinguishable from the boosting of gen AI slop?

#HanLee #AIJournalism #GenAI #AIGoldenShowers
https://leehanchung.github.io/about/

About

About page of Han Lee

Han, Not Solo

Afrois Alreadyin 23h ago

@robert this article gave me the feeling of the good old internet: an excellent, mind blowing piece that will change your perspective on a whole topic by someone you never heard of

Bluebabbler 22h ago

@robert "The UI is clean. The API is RESTful. The architecture diagram is beautiful. The outputs are wrong. Nobody checks because nobody on the team knows what correct outputs look like. They’ve never looked at the data. They’ve never computed a baseline."

piegames 21h ago

@robert the post is great, but using AI images to depict historical events is a huge red flag to the point of me not being able to trust that any other details about the story are true

SamuelJohnson 21h ago

@piegames @robert This is a better take

https://mastodon.social/@glynmoody/116373712356241202

Dr. Heather Etchevers 21h ago

@robert It would be prescient to observe the destruction of university ecosystems worldwide as a cautionary example.

Expertise was irrelevant... It satisfies the reporting requirement. It fails every real-world test. This is exactly what has happened to us over the course of (checks) my entire professional life. Optimizing to make the metric look better accelerated by metastatic reports never read again if at all.

Love: The org chart says collaboration; the incentive structure says land grab.

CaliCarol 20h ago

Absolutely fascinating and so well researched and written. Thanks for sharing.

CohenTheBlue 20h ago

@robert And the explanation is given with such artistic pathos! Similarities really hit home.

Misuse Case 18h ago

@robert @GinevraCat Someone else on a different platform shared this.

GinevraCat 18h ago

@MisuseCase @robert Was it shared without attribution? Sorry - I'm not following your intention with the comment. Can you elaborate, pse?

Misuse Case 17h ago

@GinevraCat @robert No they shared the link on a Discord server I’ve been in, it’s just funny when I see something break containment from one to the other.

GinevraCat 17h ago

@MisuseCase @robert Nice!

Iron Bug 16h ago

@robert the greatest evil of "AI" is that it eliminates the responsibility, QA and the very possibility of error reports and bugfixing. it's a slop. one can use it as it is, and nothing else. no complains, no remorse. slop is not information. slop is not code. it's just slop.

@robert Wow, over the years I've read a lot of modern Chinese history, the great leap forward, cultural revolution. All of this is spot on and hits hard.

@robert If a project needs to be done later because everyone is working on another project, leadership routinely asks, "can claude do it?" Like there are two paths to development - having engineers build it and having claude build it. And the latter is free and instant.

Every week, I am personally asked, "are you using claude? what have you done with claude this week?" (I am an engineering manager). aka, "how many sparrows have you killed?"

SAILØR 11h ago

@robert I 've never been as disgusted & enfuriated than after a few weeks into a mit online course on ai inclusion, not one actual practical iota of info on what, how, howmany specialists, what benchmarks etc, nothing but people callin each other professor evry 45sec & business "professionals" lapping up whatever fairypromise them were fed, I clocked out after handing in a paper on the use of robots to make upper echelons understand that them don't run on half da voltage as they expect people to

Chris Johnson 2h ago

@robert Well said. Great article.