Time and energy to prep meals is also a cost. I don’t know how it is in Europe, but in North America, the poor-but-employed segment of the population is often working multiple minimum wage jobs to stay afloat. Even if they know how to cook and have the tools to do so, they may be too tired when they get home to do more than pop a pizza in the oven.

Why do I still have a 32" TV? Because that’s the largest size that’s still readily available as an ordinary, cheap, flat dumb panel with a tuner. (Well, that and I don’t especially need a larger one.)

Is it terrible that I’d like to see an LLM trained exclusively on translated shoujo manga trying to give teen boys advice about this?

Yes, there are parts of Canada that remote that still have roads. I grew up in one of them. Let’s posit an urgent but not-likely-to-be-fatal medical emergency, like the torn and detached retina I had a few years ago. That required an urgent trip to a major city in particularly foul winter weather. Nearest major city to where I grew up was 800+km, and there are other towns further out than that one. Add to that battery loss in the cold, plus loss of battery capacity over time if you’ve had the car for a while, plus the vehicle having maybe already been driven that day without time to recharge completely . . . I can think of places up in that neck of the woods where I would be seriously worried that 1000km of rated range wouldn’t be enough, although it would be more than sufficient for where I’m now living.

So I’m talking about shit that, in my experience, actually happens to actual people. The segment of the population involved is, admittedly, not all that large, but it’s of nonzero size—probably on the order of a few million, worldwide, spread through a number of countries that have large areas of empty nothing.

1000km range is fucking stupid. No one should be driving that far at once

I take it you’ve never had an emergency while living in a remote area. Especially not one with cold winters that will tank your EV’s range.

Only in the US. Other countries will be able to push the prices down.

Does it do anything that isn’t in response to a human’s prompting? No? Then it can’t be conscious. Consciousness requires having a sense of self, which implies having needs and desires that one acts to fulfill without needing prompting. Even a bacterium is more conscious than these things.

Is anyone actually surprised by this? It’s one of those things that any semi-competent programmer could have told you would be the case. The study just formalizes it and adds specifics.

Thing is, that means you don’t really own the hardware that you buy, because a corporation is dictating what you can do with it even though it doesn’t belong to them. Most of us consider that unacceptable.

Pretty noticeable that Gentoo Linux doesn’t offer an option to compile OnlyOffice locally—it’s only available as a -bin package, which means that it’s precompiled by upstream. That tells me that either the available source is too incomplete to actually compile the software from, or it has some really strange licensing. Either way, it can’t be open-source software in the accepted sense.

The chain of trust starts with the owner of the hardware, not some random corporation that happens to make an OS. The owner can, if they wish, outsource the root of the chain of trust to a corporation, but that should be an active decision on their part, not something that happens just because the hardware was shipped with some random OS preloaded.

. . . And then the market will be flooded with RAM that companies preordered and can’t pay for, because the AI bubble burst before it could be manufactured.

Hey, I can dream, right? And seriously, I would be quite happy if this causes an increase in dumb appliances, devices, and cars in the meanwhile.

Actually, I’d interpret it as him losing his job in 18 months regardless of whether he succeeds or fails, since management is a white-collar job.

How concerned should I be when the documentation for complex devices coming out of China always seems to be so bad that no one except the people who designed them can program them anyway?

Telemedicine: better than nothing, already used a fair amount in the more inaccessible parts of Canada (ideally in combination with a nursing station so there’s someone with some training available to do things that absolutely need hands on location).

AI medicine: likely worse than nothing, some people are going to get killed.

They do make hardware in most of those categories, actually, but they don’t sell much of it direct to consumer in the West. And unfortunately, the way things are going, they’re going to be able to get better prices for it from the AI-entranced idiots too.

They’re really optimizing for the income of the people who make the apps. No surprise there.

And thus begins the three-legged race between imaged-based age verification and kids. (Prediction: the kids will win, but it will take the other side a looooong time to admit it.)

LLMs need to have the same warnings attached as the old psychic hotlines: “Must be 18. For entertainment only.”

That being said, I’m not sure that this is any more ridiculous than an ad asking for ten years’ experience with a piece of software that’s only existed for three. HR departments have never had much contact with reality.

With the LLM pushers driving hardware prices through the roof, will any of us be able to afford these?