RE: https://techhub.social/@Techmeme/117452462772724375
I suspect @sophieschmieg has things to say about this. (And if someone could take notes when she does that’d be great—I really need more examples of creative cursing in German)
It is *so* much fun trying to explain to my husband how OpenAI’s math “solutions” were generated and what they mean, when he fundamentally won’t wrap his head around the phrase “it can’t be that stupid”.
RE: https://sfba.social/@drahardja/117426620959164864
It’s getting to the point where the web needs directory sites because there’s otherwise no good way to find things.
Everything old is new again. Next we’re going to be putting animated “under construction” GIFs on things.
I should really sit down and write up why OpenAI and Anthropic are totally screwed if they stop training their frontier models, but only slightly less screwed for most things if they continue to train them.
This is well past “whoops, missed something” and deep into aggressive (possibly malicious) negligence.
Though if some of this shit was flagged and people were overruled so said “eh, fuck it, you want to play stupid games, you’re going to win a stupid prize”… ok, in that case I could send someone pie for having to deal with this shit.
RE: https://mastodon.social/@arstechnica/117390655190819555
Argh. AAAAARRRRGGGHHHH! This is stupid and idiotic and infuriating and if I hadn’t retired I’d be ripping multiple people to shreds at work. This is basically “what if we took 20+ years of hard won knowledge about attack mitigation provided to us by many organizations including the NSA and PLA and just… pretended that could never happen.”
All this shit in here? Google internally was hit by every single one of these issues multiple times more than a decade ago.
RE: https://mathstodon.xyz/@mjd/117326589463677193
You should hire Mark, he’s excellent at the many things he does and you will be glad that you did.
Talarico would be considered a center right candidate in a lot of the world, which says a lot about Texas. (Mostly that its enthusiastically self destructive, I guess)
We really, really need to remove the regulatory loophole for supplements. If you want to make a claim you need actual data, and efficacy studies (and long term risk studies) should be done on the things.
I’m good with monitoring studies being done by the government health agencies, that’s fine. But it needs to be done.
https://www.sciencealert.com/popular-joint-supplement-linked-to-25-higher-dementia-risk
And this is on top of half a century or more of gutting our industries and industrial development capabilities so we could exploit cheap labor and leave ourselves nearly useless.
(With those political decisions in large part driven by the capital class, this is on them too)
RE: https://mastodon.social/@cobrate/117360706453767084
This, and a whole lot of the other stuff around China’s rise in science and tech, honestly has almost nothing to do with what’s regularly blamed, and is almost entirely due to the failures and idiocies of the western capital class. If we wanted to drop a trillion dollars into pretty much any kind of science or practical engineering (instead of wasting it on AI) we could do the same. But we don’t.
RE: https://mastodonapp.uk/@brainsmatter/117357642224708355
This is something you wouldn’t normally think about — AI borrowing is so massive it’s distorting the bond markets, including for T bills. Which is, in its own way, impressive. Horrifying, but impressive.
RE: https://mastodon.scot/@dougcoulson/117343406521143514
Time for serious sanctions and bar complaints, tbh. And pushing for them to be taken seriously.
the electron runtime is several hundred megabytes. the entirety of macintosh system 7.5.5 is about 75MB, and mini vmac is 164KB. if you want to ship a cross-platform native experience, you've got options
And no, I'm not glorifying or fetishizing manual labor or anything. A lot of it actually sucks and the world would be a better place if it *was* totally automated. Regardless, it *isn't*, to a first approximation nobody's trying to, and even if we try we're decades or centuries away from a meaningful amount of full automation.
So, y'know, yay for your science fantasy view of reality? It's... nice? (Or would be if your biases weren't making it inherently fascistic nonsense) It's just not real.
Superintelligent AI isn't going to do all this for us any more than my screwdriver will. The only real difference is nobody is delusional enough to think my screwdriver could run the world, while a whole lot of choads who aren't safe to change a battery in a TV remote seem to think that Superintelligent AI can.
They're also not, even with all our tech crap, drive trucks or operate most warehouses, and they definitely aren't working mining or heavy industry or smelting plants or chemical plants. Or fixing sewer pumps when they break or cleaning up damage after a tornado or hurricane.
Yes, there's automation that reduces the need for labor in at least some of these circumstances, but that's all it is. Reduction. Fundamentally basically everything meaningful requires humans. Machines are, at best, aids.
This is one more argument, as if we need one, that everyone should spend at least six months (and maybe a year) working some kind of customer facing job. Robots aren't going to cook for you in restaurants, or serve the food or bus the tables. Nor are they paving the roads, building the houses, plowing the fields, running electrical cabling, or doing your plumbing.
One thing I've noticed, with all the AI doomerism, is just how much *contempt* all these AI bros have for actual labor, to the point where they pretend it doesn't actually exist or matter.
Like, my dude, I am absolutely sure you haven't noticed, but in the Real World:tm: actual human beings are still required for basically all productive things, and if your magic AI overlord (which is just three calvinist views of God in a trench coat) takes over then nothing's getting done.
RE: https://mstdn.social/@TechCrunch/117282410920071645
I’m sure this was totally above board and not in any way skeevy, or reflective of an organization that is skeevy from the top down.