that's why i tweeted this, but it's been a trend i've noticed for a while

my response to mandatory raytracing is this:

this is a good thing, please don't change

idk, i don't really care about the absolute frontier much anymore. it's cool that these models can do more but for the things i do it doesn't really matter other than that they'll make 5% fewer mistakes and have a nicer output to read

for most things i would absolutely trust a doctor with that, but for something where i've hyperspecialized into female/male hormones and their effects at different levels where not paying attention to it gets me on the 4mg estrogen + 300mg spiro instead of 5mg per 5 days valerate, i'm taking that

do you have a jailbreakable ps4 that you could pirate it on?

it might be insurance's fault because if you don't have Sterotypical ADHD then they won't cover it and maybe the doctor was being nice about it, idk

yeah it's absurd. i had to do a little questionnaire form before and during the actual appointment too which i guess disqualified me from it? also a 15 minute "click this button when we say X but not Y" which like, i've been playing FPS games for most of my life, i've practically trained for that

yeah i agree, with prices going up so fast though i think it'd be more reasonable to look at like steam hardware surveys and then just optimize for the most used graphics cards there

yeah that's true actually

maybe this is just a symptom of the problems here, but a lot of the docs i see nowadays are just generated with an LLM instead of being human written to the point where even if there are docs i would rather the model just read the source code to discover behaviors instead

i think nowadays if you use codex + oai models they have their own cached index of things at least to somewhat prevent this, but i do think most people just don't want their stuff scraped in the first place for whatever reason. i'm not sure what's bad about the model reading the source itself tho...

my bad for not wanting to do it as a recreational drug and instead wanting it as a replacement for not having to dance around the medical system to get the thing that will help me do anything in my life

i've tried to get a real diagnosis before but last time i was told "you just have executive dysfunction" and i was given wellbutrin, which did help a little but it stopped working after a month

looking at the dark web markets the only amphetamines i seem to be able to get are 30mg IR which sucks

er i didn't parse this correctly, i don't really get why having it read through source code for you to determine output behavior/how to use it is a bad thing

ok yeah i mostly agree with this article, LLMs at the moment are fueled by venture capitalism so they want you to use them as much as possible so that they can make money on you in the future, if you're already dependent that second part is a lot easier

i can't wait until it becomes a mundane tool

>You basically have to beg it to read human sources, it will always try to clone repos and read the source code. Keep asking it to provide you with human sources.

i've personally not experienced this, every agent i've tried has anchored it's output on the context of library source/docs...

save me p3 display that doesn't cost more than my macbook....