Yeah, it would probably take legislation forcing all of them to post and advertise prices including taxes. If everyone had to do it no retailer would be disadvantaged by being the first.
That said, I think it's a bad idea, unless retailers also have to itemize out the taxes on receipts so that consumers can see how much tax they're paying, which typically doesn't happen in Europe, as far as I've noticed (other than VAT, which is often itemized out on some purchases so that foreigners can get a VAT rebate). I think it's important that people see the taxes they pay so they can evaluate whether they think they're getting good value for their tax money. This is why I also oppose corporate taxes and any other sorts of taxes that are ultimately borne by individual taxpayers but are hidden by layers of obfuscation. Actually, there's another reason to oppose corporate taxes: Corporate taxes delegate to corporations the decision of how to allocate the cost of the taxes between customers, employees and shareholders. That allocation is an important public policy matter, and it should be decided by legislation, not by corporate bosses.
To be clear, I think there are a variety of public services that absolutely should be funded by taxpayers, and wholeheartedly support taxation for those purposes. But exactly what should be taxpayer-funded, at what level and with what efficiency are all important questions that voters should have input into, and that requires that they actually see what taxes they're paying.
GPU are retargetable (to their original use as graphic processors), but I'm not sure the same is true of TPUs and the other more specialized varieties.
Don't expect AI to ever use only a small amount of compute. You can do a lot by pre-training, but there are limits.
OTOH, I'm rather sure that the current algorithms are a lot more wasteful than a later version will be. A factor of 100 wouldn't surprise me. Personally I think the way to handle it is with a raft of Small Language Models, each one tuned to a specific context, and a higher system that switches context as appropriate. (I've seen signs in the news that we're already headed that way.)
Moore's law may be over, but the 3D version of it is just getting started. The real problem is moving the heat away from the chip. I think we're in the early part of the ramp up of 3D chips.
N.B.: That it's actually do-able was proven decades ago, but only for custom sculpted 1-off chips in a lab setting. (I believe it was the Tennessee Valley Authority...but I'm more sure about the Tennessee than about the rest.)
AFAIKT, China is 4-5 years away from "breaking into this market", if they market is the upper end of the chips. Possibly even a bit longer. OTOH, for many purposes their chips are already good enough, so they'll break into it at the lower end as soon as they have enough chips for export. (Aren't they already doing that?)
I may have a "smart phone", but I refuse to set up internet connections on it.
If they're using the enclaves built into Intel and AMD, there may be side-channel issues to deal with. ARM is closer to what Apple is trying with their enclave.
ARM's TrustZone is definitely more secure than the alternatives on Intel/AMD, but TrustZone is also subject to side-channel attacks. To a first approximation, it's impossible to run two workloads on the same CPU and keep them perfectly isolated from one another.
However, I don't think any of these secure enclave concepts are relevant in this case. The way you'd build a private AI cloud is not to run it in enclaves (which are essentially just security-focused VMs) on CPUs that are running other tasks, the way you'd do it is to devote a bunch of CPUs solely to running the private AI workloads. Then your isolation problem becomes the traditional ones of physical access control to the secure machines and securing data flowing into and out of those machines over network connections.
An open weight model is like an mp4 movie. They give you a compressed datafile and tell you to run it in your favourite (movie) player software.
Wait, did you hear that? There's a rumbling sound in my living room, it's like a vworp vworp vworp! BRB.
Well none of them really ever where. The closest we had was pre franco spain in that brief moment when the anarchists actually where doing the "workers running the economy" thing that the marxists promise but innevitably turned into turned into the state running the economy instead. Which ironically got smashed first by the communist party which did NOT like the idea of workers doing a communism without them, and finally by Franco and his fascist goons.
Microsoft can't figure out a way to market/sell this so they're resorting to astroturfing
Strange that they can't figure this out. Have they asked their AI how to solve this problem? I assume it's as simple as creating an AI to simulate an influencer, that can influence people to use AI more regularly. No brainer, really. Does Microsoft even have any interns left who could be tasked with this?
Why isn't there ever a hate button?
There is! Its up the top of the browser window, and looks like a little [x] in the tab next to the tab name
There's no sense in being precise when you don't even know what you're talking about. -- John von Neumann
