🐔 Brian Bucklew 🐔 ₑͤ＞∿＜ₑͤ ∞🌮

@FibreTigre It's a huge, expensive translation and we're doing one language as a technical test case, and turkish happened to be the winner for a variety of reasons the big one being a person had both the willingness and ability to do a turkish translation as the test case :)

caves of qud localisé en turc mais pas en français ?

Reasons to nom Caves of Qud for the Steam Awards: ▶ Loved by science fiction writers, sickos and critics alike ▶ Flexible pick, can nominate for almost every category ▶ The one the cool kids are picking ▶This is the ONLY year you can do this

the cool thing is either the ai bubble bursts and there are no jobs or the ai bubble doesnt burst and is real and there are no jobs

theres lots of fancier fighter jet devs, I'm not so fancy, just an aging workhorse F15 that can tow 14k bl bomb payloads day in and day out reliably

What we needed: A useful index of all human knowledge, easily usable to go right to the source. What we got: A medicore redditor that read every wikipedia page with marginal comprehension and really wants you to like them.

we'll end up with 2, because of monopoly laws; but any AI megacenter > 2 is going to end up as like a monkey habitat or something

like even if LLMS work bigly one problem is we need at most *1* datacenter stuffed full of every information we've ever made, every other copy is redunant and will be ripped out as soon as the tech finally matures (which it probably already did 18 months ago)

every vet website: your dog should not eat [random thng] because it may cause gastrointestinal upset my dog: i ate the broken pieces of a plate that used to have catfood on it and i feel nothing my stomach eats the edges off the cermaic this is so easy true story from 20y ago

I dunno how you can disbelieve in actual real life spirit gods when you can easily look out and see how the capital egregore so easily infests and puppets so many people.

I really don't know why we're just letting the capital egregore run society. My brothers and sisters I have to tell you - it is really pretty bad at it.

Imagine if we took all the funding we gave to AI for one year and gave it to someone like dolphin speech experts or primatologists or blacksmiths or something. Like just try it and see what happens

this answer is objectively wrong

like even if you view it as a statistical model, I'm sorry but statistically your idea is a bad one, in the space of all ideas

AI increases code velocity by 1.02 and tech debt velocity by 1.8 this is good because

It is cool that we have a nice solution for autonomic cerebellar functions when we get a good executive built though.

Maybe we get there, but I really personally see no signs at all of anyone approaching the problem in ways that would actually solve it; instead they're windmill slamming dumb loops on top of LLMs. These are standins for real executive systems we're decades from mastering imo.

AI coding makes you think it's way better than it is, because it can stand up kinda working hobby examples or demos really fast, but the step from 90% working to production software you can actually ship and support is a vast abyss. Completely insurmountable for current gen ai

Nobody's even made the very very start of a proof of concept for a slightly good one.