gerikson

This piece, although in a way defeatist, also gives me hope because there's at least one other person who has the same general feeling about LLMs that I do, and is a better writer.

https://blog.glyph.im/2025/06/i-think-im-done-thinking-about-genai-for-now.html

I'm gonna think that the latest drumbeat of pro-LLM posts (tpacek's screed, this excrescense) is a last gasp of a system running in midair like the Coyote, before the VC money dries up.

Lobsters went down a VC financing rabbit hole the other day (thanks to me and @dgerard) and a user horked up this absolutely bonkers defense of OpenAI losing a galactic sum of money:

https://lobste.rs/s/wjb9ox/minio_removes_web_ui_features_from#c_rgatzz

(reproduced below in case it is removed in shame)

OpenAI is very different. They mainly lose money on ChatGPT, but it’s not really lost money, because they in turn accumulate fresh daha to further train their models. Data that none of their competitors have access to.

OpenAI is also different because AI is a major geopolitical factor at the moment and unless you’ve been living in a cave lately, you must have noticed that geopolitics is much more important than money these days. ChatGPT is an incredible intelligence gathering channel and cutting access to AI APIs would make US sanctions hurt that much more. The only other country that can compete with US companies when it comes to bulk training data access is China, via their social media alternatives like TikTok and RedNote. You can imagine the geopolitical implications of that too.

Maybe... like I mentioned, Nokia's S60 application stack was a mess. The underlying phone software and platform might have been there, but the 3rd party ecosystem wasn't. This was a huge part of the success of the iPhone, that 3rd party developers had a stable platform to develop for, and a steady financial partner (Apple) paying them.

No offense against Nokia but I really don't think the company had the mentality to offer that.

Nokia had great hardware, but crappy software (and I say that as a heavy Series 60 user back in the day). In a parallel world, Windows Mobile could have ridden that hardware to a glorious future, but it was transparent that Elop's acquisition was just part of a Byzantine internal Microsoft play.

Good luck! I'm rooting for you.

I hate I'm so terminally online I found out about the rumor that Musk and Stephen Miller's wife are bumping uglies through a horrorfic parody account

https://mastodon.social/@bitterkarella@sfba.social/114593332907413196

why is robot lady in queue with springy legs so darn sexy

seriously though, good piece. Hopefully the fever breaks soon (yeah right)

Anecdata: if you're working in IT in Sweden you can get away with just English. I know a guy living in Berlin who hasn't ~~bothered~~got around to learning German yet, he also manages with English.

But it's a big Community and different parts have different requirements and of course different expectations.

LWer asks: "what if property-based suffrage, but with crypto?"

https://www.lesswrong.com/posts/wsBQGZfJcLEshp5cd/beyond-democracy-a-system-where-citizens-vote-with-their

Oh man this is peak venture capitalism crossed with Factorio - valuations are actually cash, and a factory is a black box where you just upload new software and other stuff comes out.

Let's take your average holder of car manufacturer stock. You're holding the stock because you believe the car manufacturer will continue making competitive products, and you'll get either dividends or higher valuations. Then OpenAI pitches up and offers you - what? They don't even have stock! Even if they did, you're exchanging a stake in something known for stake in an enterprise that have never made any cars, and when asked what kind of business plan they have they look shifty. No fucking way anyone will sell their stake for less than double what they have, especially if they find out the factory they're selling is gonna produce machines that will kill us all.

Wow he looks even dorkier in video than in photos.

Stupid sexy robot judge...

One thing an adversarial judicial system like the American one is that if one party sloppily use GenAI to write their documents, they can lose, because the other party can point that out. A lot of the excuses to use LLMs in software development is that modern software development is terrible anyway, so if you can get your slop to market faster than some other schlub, you probably won't lose customers. When there's a balanced incentive to point out hallucinations, they (hopefully) won't get that far.