Claude outage

If LLM use were as valuable as the adherents claim it is, this news would be on par with AWS US East 1 being down.

LLMs neither have the mechanical reliability we expect from computers (does it the same way every time) nor the flexible reliability we expect from biological intelligences (solves or works around the unexpected sub-problems as they arise).

If all of Amazon went down in 1998 or 1999 it wouldn’t have been news on par with AWS in 2025.

What a weird top comment. It opens with some subtle name calling ("adherents"?), an unwarranted conclusion and then a statement about what LLMs lack (which is no surprise) that we're meant to draw a conclusion from?

I find the debate about LLMs rather exhausting. I find them useful and almost every day someone on social media tells me I'm mistaken, lying or merely deluded .

It also happened during the night/non-working hours for people in NA so that has something to do with it.

It is very easy to switch. Today I got things done with ChatGPT. If I didn’t have any LLM available only then it would be disaster.

Or you are working on easy things, or my favorite incompetent.

I'm actually working on something moderately hard but I am admittedly slightly incompetent.

Life is much easier when you assume the worst when you don’t understand why someone is doing something.