It would drastically affect most of Northern Europe, also the Eastern US. But Iceland would probably be the most seriously affected. Greenland might refreeze, I believe that the "little ice age" was responsible for the failure of the Norse Colony on the shores of Greenland.

If you read to "popular science news", you'll get a(n oversimplified) version of the news. More exact sources are available, often all the way to the raw data. But they take more effort to read.

You seem to think all robots are the same. We've got LOTS of different kinds of robots, from robot pencil sharpeners on up. We've got robot forklifts, robot snakes, robot airplanes, etc. Humanoid robots are just another kind, but a kind that's potentially quite useful in environments shaped for humans to operate in.

Whatever, it correct anyway. The press tends to be xenophobic. They'll criticize France or Canada just as readily, though less fervently. For some reason they tend to go easy on Mexico.

You;re assuming they care whether it "musician" is an AI or not. Usually they won't. Think of it as "mood music"...it's purpose is to establish or reinforce a particular mood.

Actually, it doesn't matter much whether the "musician" can feel the emotion or not...if it did jukeboxes would never have been a thing. What matters is that it establishes a particular mood in the listeners, and note that that was a plural.

But isn't that's what popular music is reported to be already? If only a formula can win, then an AI should be able to do it.

FWIW, my tastes are "folk music" and sometimes "filk music". That's much less a formula, but possibly only because the sample size is small.

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.

The new stuff is under copyright.

In addition, the part of that money spent on computer centers will be useful even if AI doesn't pan out. It's not like investing in tulip bulbs. If AI doesn't pan out, it will just take a few years longer to pay for itself.

That said, AI will pan out. Even if there's no further development (HAH!) the current AIs will find an immense number of uses. It may well be "growing too fast", but that's not the same as worthless. (But expect well over half of the AI projects that are adopted in the next few years to fail. People don't yet understand the strengths and weaknesses. Unless, of course, AGI is actually developed. Then all bets are off because we REALLY don't understand what that woud result in.)

It's going to take more than one more efficient algorithm. OTOH, there've already been improvements in more than one algorithm. Nobody knows how far that could go, but the best evidence is that it could get a LOT more efficient. (Consider the power usage of a human brain...it uses a lot of power for an organ, but not really all that much.)

I'm guessing this is a summary:
Banks are legally allowed to loan more money than they have in deposits...to a degree. They've occasionally been found to go well beyond that limit. And they aren't carefully audited often enough.

Whether that's an accurate summary or not, it's true, if a bit shy on details. (I don't know the details this decade. But there probably haven't been any basic changes in the last few decades.)