humans supervising agents, not dark factory where agents supervise agents. providing a very high degree of automation, not fully automated. controls implemented using deterministic tools, not via markdown prayers.
opposite of 'lights-out', which in factory terms means a factory that's sufficiently automated that it can operate most of the time with the lights turned off because there's no people inside.
It is becoming common nomenclature. I think I first heard about Dark Factory / "lights-off" in AI about a year ago now.
There's a similar concept in food delivery, where restaurants that only deliver food (eg Doordash / Uber Eats only) and don't have a physical store for customers are called Dark Kitchens. They try to automate as much of the food preparation as well.
I know plenty of people who like to create—but who also have a better technical understanding of LLMs—who use LLMs in their workflows (some even use LLMs finetune LLMs on their own work and then incorporate them into their workflows.)
Most people who are non-technical (including most creators) have an extremely naive view of what LLMs are, mostly driven by what the media, and shills who are mostly targeting audiences that aren't creative are focused on, and their response to LLMs is shaped by that.
I should have said "the people who like to create don't use LLMs for the parts they like creating". I like making products that are easy to use and useful, I have LLMs write 100% of the code but I still do all the UX by hand, because that's what I enjoy.
The name makes sense because Aluminium has an -ium suffix like Chromium. There's also no reason for the project name to agree with the US pronunciation of the element.
Well, it makes sense and it doesn't because it makes it sound like this is a 'lightweight' version of the Chromium-based products while the opposite seems to be true. Call it Osmium instead, that's got '-ium' and some weight to it just like this thing.
My dad always pronounced it a-luna-min, so my whole life I thought that there were 3 pronunciations, and the fact that there are only two correct ones feels strange to me. Not sure where he got that from, maybe he had special metal from the moon.
online writing before 2022 is the low-background steel of the information age. now these models will all be training on their own output. what will the consequences be of this?
I love how we have such a poor model of how LLMs work (or more aptly don't work) that we are developing an entire alchemical practice around them. Definitely seems healthy for the industry and the species.
reply