News at 11.

Or "profits beyond reasonable repayment for capital, indicative of monopolistic, or monopsonistic, power over the other trade party".

I think we need to codify what "normal" is, and what should happen otherwise. Winners won't willingly give up their winning lottery ticket unless forced.

It isn't like Micron made any amazing technology leap, or strategic business decision that paid off. No, they're lucky. And those in power have decided that AI companies win.

They obviously failed their duties to direct everyone else.

Seems like plenty of people at NVIDIA would work harder not to let their products get into the wrong hands if they would feel pain when they failed.

Dell who couldn't identify the product before Apple, decides they can solve the same problem how they always have. Good luck with that Dell.

The only way the "old guard" tech are still around is by momentum and (customer) memory loss.

as they want to believe. They're just the psychopath in charge. The front man.

Surely by now.

Those CPU extensions for creating a separate area that is secure. The ones Intel removed from consumer CPU's which broke playback of 4K bluerays on desktop PC's from that point forward.

Meaning partner, children, friends, etc... can all hear information ahead of time too. Just think Nancy Pelosi's husband who seemed to beat the market reliably.

It'd certainly help everyone move to versions that are secure if the insecure versions would turn themselves off. Meaning the creator/publisher would remove approval for easily broken/hacked versions, and after a few warnings would disable/block the broken versions from running at all.

Yes, I'm aware of issues with this pattern, but it still sounds "more secure" than what's typical today (relying on each separate developer to decide to upgrade).

Be honest about the purpose, and measure if you reach those goals.

Public school is "free" babysitting. Indoctrination. Forced "socialization" (break their wills). All to create useful cogs. Serfs.

Oh, and keep our mental health medical providers in business with plenty of early trauma and victim blaming.

"The 26B Mixture of Experts model activates only 3.8 billion of its 26 billion parameters in inference mode, giving it much higher tokens-per-second than similarly sized models."

Isn't that a 3.8 billion parameter model then? Created from the 26 billion version. Or do they mean it "mostly" sticks to 3.8 billion parameters.

That's the other side of the coin that some people likely don't want to happen.

And isn't insurance kind of like betting or gambling on an outcome eventually happening? Rather than a specific upcoming moment's result, the fact a result is likely to happen or not.

It should be publishable monthly.

Less frequent information is not better for the public. It's better for the companies.

Can't remember who's phone finally did this (OPPO?), but they'd add a hash to recorded images captured with the device. To prove this is what was recorded from the sensors, not a generated or modified data file. Guessing at the firmware level to try to make it harder to fake, and depending on how it was set up you might even be able to confirm which device captured it (would be nice if you had to already know the source ID to compare against a device, with yes/no answers).

Yes, there are issues to iron out before it could be a generally assumed useful tool. But trying to "identify what is real from the beginning" seems like a better solution than "finding what might be fake" (some of the time, after some reputation damage was already done, and sometimes creating false positives marking real images/videos as fake).

Then we could create methods to connect the original source material (for measured/recorded authentication) with any customized versions (cropped, filtered, etc). Maybe even with a numerical estimate for how likely this is a likely starting point for the result file, and how it was created (to recreate it by other people if they don't want to just trust the source).