Solar isn't free. When it makes sense, it gets used in data centers: https://seia.org/blog/ai-leade...

If they actually do invent general AI, then they WILL make their money back. To get there, they need new algorithms, not bigger data centers.

It's gotten a bit confused over the years. At one time, an average student could expect a C. Calling someone a C student wouldn't have been much of an insult. A D student was under-performing. The real shame was at the point of D-.

But then, starting in grade school, parents expected A's and B's even from objectively average students. A's were for students who might even be at a point where they might skip a grade. Kids started betting grounded for too many C's and eventually for any Cs at all.

That's not to say it was all perfect. There was a certain un-fairness to grading on a curve per class rather than over an established history of classes. An average student who found themselves in an exceptionally bright class might get a D or an F in spite of being objectively average. The F being especially unfair as a small change in fate might have gotten them a C or even a B for the same performance if they happened to be in a more average class or even below average class. It just wasn't that good of a predictor for later real world performance.

A million little girls want a pony.

Right, in other words, third world bargaining tactics.

Apparently they have 350,000 workers in corporate, so firing 14k is about 4% of their total employees. Not a huge cut.

If you are a manager, and hire five managers below you, then it increases your importance. The optimal strategy is to hire as many people as you can. If they are useless, that is a bonus (because then you have to hire more people to get the job done).

According to IEEE, they are using jet engines to power data centers because they can't get grid connected hookups. Connecting to the grid is apparently cheaper, and they will do so as soon as the power is available.

Then we will have "dark AI datacenters" to match all the "dark fiber" that was laid in the .com era.

I was not aware that modern data centers use jet engines. That is new info to me.

It seems like any AI data center would be ok with 95% uptime. If you can't ask questions for a few minutes, that's ok. Build the datacenter, then bring in the generators when they become available.

Order of operations happen deterministically.

Thanks for the info.

I make a few reasonable inferences that don't involve magic. Meanwhile you invent active shooters with no shots fired and no gun present. I'm pretty sure they didn't use divining rods or the magic 8-ball to locate the kid in question. The picture was obviously available (since it is documented that it was shown to the kid). It's documented that the kid was cuffed and on his knees. It's documented that there never was a gun.

The "gun" turning out to be Doritos must have been fairly obvious since when the kid pointed it out, he was not arrested and taken away.

It's also obvious that had they looked at the picture FIRST, the kid wouldn't have even known they were there (nor would anyone but the principal and perhaps a few others in the office).

You can feel free to lick as many boots as you like, but leave me out of it. If the police want respect, they'd best get busy earning it.

So you figure they were waving ostrich feathers at him threatening to tickle?

Do you think the only actual guns in the school at the time didn't loom large in the kid's vision?

How do you suppose they picked out the "right" student to hassle without a picture?

In the before time when I was in high school, searching for weapons was a job for the unarmed principal or the football coach. Either they had the picture on them or they held the kid WAY longer than necessary after the pat down revealed nothing.

The parents should sue the crap out of the cops, the school, and probably the manufacturer of the scanner.

Agreed, the actual publication is considerably more ambiguous. In my case, 2K really would be about the max, but I can easily see other people would have a good case for 4K.

Though notably they're talking about ability to discern the difference in an A/B test. A somewhat lesser display may still make no difference in the 'experience' of normal watching, but that would require a follow-up study.