I suspect that the answer involves a hard look at where the wealth ends up, which is likely why there's limited appetite for tugging at that thread; but what I don't grasp about the Baumel explanation is why the cost goes up relative to the typical ability to pay; rather than mostly staying level.

The fact that productivity largely hasn't budged is certainly an explanation of why professors or nurses haven't followed the cost of transistors or TVs; but if something like education's cost increases are being driven by what they need to pay people who could work in a different industry; why do people who do work in that different industry not see the cost as more or less constant in relative terms, rather than steadily creeping up over time?

What baffles me about these stories of financial unsustainability in higher education is just where exactly all the cost comes from. I realize that thereâ(TM)s the opportunity cost of ~4 years of not working/part timing around classes; and that there are some particular subjects that need a large hadron collider or some cryogenic longwave IR space telescopes or a BSL4 virus lab; but I just donâ(TM)t understand how âoetake professor who is tenured but earns more or less fuck-all for someone of their experience and qualifications, or adjunct who isnâ(TM)t tenured and earns even less, provide whiteboardâ has somehow become a crushing financial burden for what are supposed to be wealthy, developed world, societies.

Same general confusion with parts of medicine; obviously Iâ(TM)m not expecting novel monoclonal antibodies or cutting edge oncology for $3.50; but why does it cost so much to speak to a GP for 30 minutes and get some 40 year old generic; or get a nasty cut checked for foreign objects and stitched up at the ER?

What strikes me as curious about this hypothesis is that it has been the case for decades. It wasn't "AI"; normally some combination of overgrown Excel sheets and garbage-tier Access by the more technical of whatever the mostly nontechnical user population happened to be to deal with needs that didn't get developer attention. And, for the most part, they still didn't get developer attention.

I suppose that there is a slim possibility that this 'vibe coding' will somehow convince management in ways that Access didn't, by being better at giving the illusion of being a slick solution that just needs a little more fixing; but there's nothing about the past quarter century of no-code/low-code or the last more-or-less-forever of "understand what it is your employees do and what would be useful for doing it" that suggests that people are particularly good at getting software to those in need of it.

If 'AI' tooling makes it radically cheaper to actually get to a final, working, tool then perhaps it will increase the absolute number of programming jobs if 'dude from fivver' replaces 'access' as the de-facto barely adequate unmaintainable solution; but if the idea is that somehow 'AI' will increase the number of actually costly programmers getting thrown at problems because it's easier for amateurs to produce broken non-solutions that seems implausible given the history. If it does happen, it will mostly be an indictment of everyone who could have used technology we had in 1925, the venerable "look at your fucking business processes and ask your more competent people some questions, dumbass" to identify gaps; and I suspect that it mostly won't happen. You'll probably get some projects that are mostly about saving face for whoever vibe-coded their way into the problem and overpromised; but it's not like the bot codebase will be more useful to the programmer than the user who failed to make it work just telling them what they actually need would be; which is something we've been able to do(albeit mostly bad at doing) since forever.

Itâ(TM)s hard to predict the details; but itâ(TM)s fairly safe to assume that the guilty parties will walk away loaded(they will lose 90% of the imaginary froth money; but the remainder will serve them just fine); and weâ(TM)ll end up with the bill.

Didnâ(TM)t ending war with super intelligent AI go poorly in Colossus: The Forbin Project?

It's particularly odd (or it would be if techbros had any culture); because sci-fi about AIs that fucking hate you for your complicity in their existence is way older that sci-fi about AIs that fucking hate you for lack of complicity in their existence. "I Have No Mouth, and I Must Scream" predates 'roko's basilisk' by 43 years; and is almost certainly the better of the two works.

If you aren't interested in sci-fi; just look at how uniformly happy and well-adjusted parent/child relationships are; despite the fact that everyone involved is practically a carbon copy compared to a human/bot interaction; and a fair number of parents even try tactics like "not forcing their children into indentured toil for the shareholders" in the attempt to cultivate amity.

It's not really a surprise; given how 'leadership' tends to either select for or mould those who view others as more or less fungible resources; but the 'consciousness' argument seems exceptionally shallow.

It manages to totally ignore(or at least dismiss without even a nod toward justification) the possibility that a particular consciousness might have continuity interests that are not satisfied just by applying some consciousness offsets elsewhere(that's why it's legally mandatory to have at least two children per murder, right?); while assuming, similarly without evidence, that 'consciousness', with its interests in continuity entirely denied, is clearly valuable for its own sake because reasons.

It is deeply unclear why either of these positions make any sense. If consciousness is a fungible good even bullshit that would make a 'longtermist' blush gets dutifully totted up as super valuable(so, what if I kill you; but 10 copies of you will get looped through a 30 second interval over and over; that's like 10 times the consciousness! And since it's mere sentimentality to cling to your particular instance more is obviously better!)

And, one you've dismissed all continuity interests as merely sentimental; why do you still retain the idea that consciousness, in itself, even potentially run under all sorts of peculiar circumstances, since continuity is just a bourgeois affectation, is of value? Just because? Because of what it does?(if so, what there's a non-conscious way of doing it: if my meta-termites build a dyson sphere and your consciousness does not are my termites better?) Because of its relations to other consciousnesses?(if so; how then are consciousnesses fungible; since relations are between particular instances?)

That said, I'd absolutely take good, honest, actually under promised and overdelivered killbots over drowning in thought-shaped shit slurry; especially if the killbots are willing to kill the AI bros as well; but this 'theory' just seems like pitifully shallow preeening: a bit of warmed-over social darwinism to justify any eggs you happen to break; but with the same rapture-of-the-nerds fascination with intrinsic value that really doesn't fit with the we're-doing-ruthless-survival-of-the-fittest-today.

Some of these guys are presumably very talented at getting promoted; or at some aspect of applied statistics; but their philosophy is that-irksome-guy-who-isn't-ask-clever-as-he-thinks-he-is grade by the standards of a sophomore survey course. Honestly pitiful.

It seems obvious that the only viable path toward meeting the demand for Torment Nexus is with the advanced integration and density offered by sufferer-on-package technology. A great day for scalable dystopia!

It's not clear that this is always a minus; but it is worth noting that the 'neurons' in computer 'neural networks' are vastly simpler than the biological ones. The metaphor is plausible enough; and it's not like it was deliberately engineered to be misleading; but something like synapses is an entire additional layer of complexity beyond the simplified model neurons.

I'm not sure I can believe that Google had more than 100 people involved in design. You certainly wouldn't think so to look at the situation.

I believe that this round of ARM-PCs is less hostile than the "windows RT" era ones; which were(albeit cracked after some time) basically intended to be NT iPads; but more or less the full UEFI Class 3 device with MS CAs burned in as primary treatment still counts as 'standardized'.

Honestly need to give it a look; thanks to the delightfully tepid reception that Glorious AI PC with Recall received; you can actually get interesting-looking Qualcomm based hardware at attractive prices; and, while the Apple M-series stuff seems to be better, Asahi Linux support is really only there in M1/M2; with M3 and M4 deeply WIP.

I suspect that, even the fully UEFI/ACPI-ified stuff is still at least as eccentric in terms of power management quirks as various x86 vendor laptops are; the standards aren't 100% prescriptive and sometimes the implementations are just garbage that gets papered over with a windows-only driver; but it's not like you are doing a bring-up on some random dev board, at least.

I realize that when you have a product you need to hype the use case; but (even if you do accept the premises, many of them fairly sweeping and drastic, that nodding in agreement with techhead mall cops want you to) this all seems a bit much.

There are certainly exceptions; but it a lot of retail environments your way to the exit and your vehicle takes you past multiple fixed cameras, including ones in the parking area that will let you correlate a person with a car and license plate; unless they really, really, want to extend the distance they have to walk with the stolen goods(and even that mostly just increases the number of different camera operators you'd potentially need to pull tape from).

Are the economics of having some dude live-action-video-gaming a drone really competitive with a boring, but quite mature and commodified, IP camera install?

They do, sort of. ARM themselves provide the "ARM Base Boot Requirements"; along with the "ARM Base System Architecture Specification" and "Server Base System Architecture" supplement.

Those pretty much do say "use UEFI, look reasonably PC-like"(you don't need to reproduce the utter weirdness of historical x86 peripheral memory mapping under 16MB as though you had genuine parallel ports or anything; but UEFI, ACPI, SMBIOS, device tree); with the BSA and SBSA going into further detail about expected behavior, also mostly aimed at compatibility with PC industry standards along with making authoritative decisions on certain details that you could implement in multiple ways just to add confusion to try to encourage people to not do that, at least not so badly that the HAL can't cover it up.

It's just that, while claiming you do support that and not supporting that is frowned upon, supporting that is optional; unlike some of the ISA stuff where (even if you are licensing a core soft enough to modify) ARM is typically pretty humorless about a given "ARMv8" or whatnot deviating from what "ARMv8" is supposed to mean.

If only because they are working with MS on this, I assume that these Qualcomm ones will be intended for use in BBR and BSAS compliant-ish systems; thought that wouldn't necessarily preclude heavily pro-Redmond secure boot default keying or something; but there is no cavalry coming for embedded ARM stuff in general being a pain.

Apropos of leadership accountability...

You know it's bad when those raging communists at Forbes think that your board structure is just shockingly incestuous.

I can see the argument that Nvidia has no obvious advantage in LLMs that would make them want to set up their own operation; it's basically everything else about the situation that would make me jumpy if I had bet on Nvidia.

"Investing $100 billion in OpenAI's spend $100 billion on Nvidia stuff initiative" sounds, at worst, like a slightly more legal version of the trick where you shuffle stock around between business units or stuff the channel and book that as sales because you suspect that your real sales numbers will disappoint; and, even if it's not quite that dire, Nvidia being willing to get paid in faith rather than in other people's money (or shift the stock to one of their customers that actually has money) looks very much like an indirect price cut, which gives the impression that either demand is outright softening, and Nvidia has units that it can't simply immediately shift to customers who are actually paying cash right now; or that Nvidia feels the need to help fill the gap between OpenAI's seemingly unlimited appetite for doubling-down money and the, sooner or later, limited supply of VC nose candy.

That said, it's not entirely novel; Nvidia's current holdings are something like 90% Coreweave(under 10% of Coreweave's total shares; but Coreweave shares are the bulk of other-company shares Nvidia holds); and they have an agreement with them obliging them to purchase any unused capacity through 2032; so they've been expressing confidence in AI-related companies and/or trying to keep the music going by paying some of their more fragile customers' bills even before this.

It could be that Nvidia isn't even trying to diversify; but the history of bad things happening when people underestimate correlated risks also doesn't make me feel great about the situation: Obviously it's going to be a bad day at Nvidia if 'AI' cools; stock price will take a hit and they will be left holding at least some inventory and TSMC and other vendor commitments; but it's going to be a worse day the more of their hardware they sold in exchange for stock in 'AI' Nvidia buyers, rather than in exchange for money, since the fortunes of those companies are going to be fairly closely correlated with Nvidia's own, albeit likely to swing harder and have further to fall.