I mean, we’re watching it happen. I don’t think it’s hypothetical anymore.

It’s already happening. GPT 5.2 is noticeably worse than previous versions.

It’s called model collapse.

Oh, the guy from Hermit Tech! He’s great, and his blog is hilarious and poignant in turns (though sometimes both at the same time).

You wanna know who really bags on LLMs? Actual AI developers. I work with some, and you’ve never heard someone shit all over this garbage like someone who works with neural networks for a living.

I won’t link directly to the video, because it does show an extremely violent brawl (and a man’s final conscious moments; to be clear, I’m not celebrating his death and never would). But the Wikipedia article on it links to news sites and analysis, and several of those have the video.

https://en.wikipedia.org/wiki/Killing_of_Quentin_Deranque

I don’t see how this could even be in question. The sequence of events in the video is absolutely plain as day. Nazis picked a fight and got their heads kicked in for it, case fucking closed.

He’s never gonna be their Charlie Kirk, as if that would even be something to aspire to.

They’re also not providing a large language model, so they actually did have a path to profitability. It’s keeping LLMs updated and running that costs so much money that companies trying to do so are losing billions, and Midjourney doesn’t have that problem.

It’s just that their path to profitability was built on plagiarism on an astonishing scale. You’re spot on, they should have been utterly destroyed right at the start.

Dare I ask… Is that real?

Ugh. I hate what we’ve become.

And it’s not even working. Not one of the AI companies is profitable. So they’re putting the hope for profits some time in the future over sanity and safety.

I have a work-supplied laptop with Windows on it. I use it maybe once or twice a month, just for the things requiring a VPN. The rest of the time it sits there gathering dust while I get real work done on my Linux laptop.

The specs on the work laptop say it should be a performance beast, but my Linux machine (with half the RAM) runs circles around it.

I’m watching that happen in my industry (software development). There’s this massive pressure campaign by damn near everyone’s employers in software dev to use LLM tools.

It’s causing developers to churn out terrible, fragile, unmaintainable code at a breakneck pace, while they’re actively forgetting how to code for themselves.

A while back, I was thinking about upgrading my living room entertainment PC. It’s got a decent video card in it, but some of the other hardware is getting long in the tooth.

Now, my plan is to focus on software tweaks to squeeze the absolute best performance I can out of it, and keep the hardware as-is until it starts physically breaking down. And when that happens, I’ll find refurbished hardware to upgrade it with, rather than spending the exorbitant fees to buy anything new.

What mystifies me about all this is that it’s obvious what the end goal is: No more PCs, and everyone just rents dumb terminals connected to AI data centers that run everything and have all the compute power. The problem is that literally no one but AI companies want that. Not consumers, and not other companies that sell software and services to consumers.

When cars replaced carriages, it was because people actually wanted them. Cars had real-world benefits over horses. But this shit? No one wants it. Gamers want game performance you simply can’t get with streamed games. People who work with computers for a living don’t want their ability to do anything to vanish if their ISP has an outage.

Shit’s gonna get stupid, fast.

???

We can’t vote with anything but our wallets here, so this is a thing we can do to at least reduce their income. And considering they’re losing close to $100 million every day already, I can’t help but suspect nickel and diming them a bit still hurts.

I wonder how OpenAI’s investors feel about that expense, considering they’re losing somewhere around $80 million every single day.

At the beginning of the year, they projected a $14 billion loss for 2026. They’ll almost certainly exceed that by quite a lot, and may have lost $12 billion in the previous fiscal quarter.

They aren’t bleeding money. They’re a veritable cash volcano, blasting it into near-Earth orbit.

At this point, I wouldn’t be surprised if they’re working on it. He’s destroying their bottom lines.

That said, if you go after the king, you’d best not miss.

Neither does Trump.

I would like to subscribe to your newsletter.

The AI boom

They misspelled “bubble.” None of the AI providers have a path towards profitability.

Yes, it’s base64. And what’s behind it could be anything that can be attached to an email.

In this case, it’s a PDF. If the base64 text can be extracted accurately, then the PDF that was attached to the email can be recreated.

The challenge is basically twofold:

1. There’s a lot of text, and it needs to be extracted perfectly. Even one character being wrong corrupts it and makes it impossible to decode.
2. As the article points out, there are lots of visual problems with the encoded text, including the shitty font it’s displayed with, which makes automating the extraction damn near impossible. OCR is very good these days, but this is kind of a perfect example of text that it has trouble with.

As for my approach, I’m basically just slowly and painstakingly running several OCR tools on small bits at a time, merging the resulting outputs, and doing my best to correct mistakes manually.