froztbyte

in today's world? too many possible answers to that question

some UN-associated ACM talk I was listening to recently had someone cite a number at (iirc) ~~$1.5tn total estimated investment~~ $800b[0]. haven't gotten to fact-check it but there's a number of parts of that talk I wish to write up and make more known

one of the people in it made some entirely AGI-pilled comments, and it's quite concerning

this talk; looks like video is finally up on youtube too (at the time I yanked it by pcap-ing a zoom playout session - turns out zoom recordings are hella aggressive about not being shared)

the question I asked was:

To Csaba (the current speaker): it seems that a lot of the current work you're engaged in is done presuming that AGI is a certainty. what modelling you have you done without that presumption?

response is about here

[0] edited for correctness; forget where I saw the >$1.5t number

tomorrow

read to the tune of, naturally

yep exactly - the plant alone is only part of the infra spend, and there's so much else that matters too

it's just such a fucking terrible way of doing this, and it sucks

imagine how fucking terrible it must be to be in this room

(and I won't lie: there's definitely a moment that Inglorious Basterds briefly flashed to mind)

ADQ any% speedruns

(A = assholes)

has to? sigh

it makes me so deeply fucking depressed knowing how much near-abandon infrastructure spending is happening. minor win is that power plants can work for other things too, but it still runs headlong into a pile of other issues (transmission network, generation method side effects, etc etc)

holy shit, across 3 comments you did a full distributed darvo

stellar example of shitheadery so early on a sunday!

oh my dearie me, I shall have to clutch my motherfuckin pearls

so it's been observed by many that github's been getting worse for a while as they keep shoving copilot into every corner

with the upgrade diff review, I wanted to quickly fold closed the 485 files in the diff. I could've sworn github's diff view used[0] to have a button for this, and I know bitbucket does[1], but nope. so of course I open browser inspector to dig at elements (then quickly iterate over them with .click() in the js console)

which is when I noticed that even the elements are renamed for copilot:

<copilot-diff-entry data-file-path=".cargo/config.toml">

which both makes me wonder my memory is right and this did used to have a button that was just overlooked in the rush for terrible chatbot shit, and makes me boggle at how astoundingly far the org is deepthroating this nonsense

[0] it's been a few years of no longer actively using github

[1] fairly recently for client work

good lord, I barely get to look at the sub for a few weeks and a whole fresh new era of people too short for the ride show up