>bitching about a working fusion reactor specifically designed to make specific isotopes instead of power for your permanently attached phone that you are addicted to

What's next, gonna complain that your bulldozer isn't winning the indy500?

It's even worse than that. They are talking about risk chances. It's literally clickbait, and I'm not convinced the journalists these days even understand how they are fucking up and misleading. It may actually be stupidity and not malice...

So say it IS an 11 percent risk increase for X. Eating that occasional hot dog then makes your original risk for X increase BY 11% not an 11% risk.

Say for the heart attack: at your age group lets just say your baseline risk is 0.35% of you having a heart attack and kicking the bucket per year. Increase that original 0.35% by 11% and you get.... a 0.388% chance.

Pretty sure a delicious belly burner every once in a while is worth a whole lot more than the 0.03% higher chance of you blowing a heart valve at 60.

Let's see:

Hmm. All of them. They require the user to have more than two brain cells to rub together to spark a thought, but there are tons of options. Some are integrated to Desktops / Package Managers and can be turned on to do automatic work too.

1: BTRFS / ZFS snapshots.
2: DD images
3: Simple tarballs
4: Haven't found a non-embedded Linux yet that doesn't support disk / partition cloning software, especially sparse file disk cloning so you clone only the actual data.
5: Did I mention FS snapshots yet? They are made specifically for this. And some are Enterprise Ready level of support.
6: Some package managers can be set up to be able to revert changes to pre update / install states too.

Any / all of these ways can be used with ANY age. Haven't changed much on that ZFS cold storage volume that you just fat fingered changing the permissions on? Who the fuck cares if the snapshot is 180 days old? Take a new snapshot, restore the old snapshot, and compare file differences sorted by something other than the fucked up perms... It's not rocket surgery.

Or store a snapshot just before you do any major changes like perms. Then it's an easy rollback to 5 mins ago. But you still have the option to use the 180 day old snapshot too.

Funny, haven't seen an ad in 11 Pro. Same install since I got and tossed new parts in this PC... just after 11 came out. I can also turn updates completely off, or to only notify me when they are available. Kind of like I have absolute control over the "hidden" settings. Like GPE. Or the Registry.

It's almost like the "I'm a cheap bastard and buy the absolute cheapest shit I can" (maybe, haven't seen it myself in Home on my dads laptop I have to admin), and "free" versions get ads.

URL and QR code that takes you somewhere remotely relevant? I wish. Shit sold "Solar Ready", with QR codes that take you to the manufacturers site to try and sell you their Solar Garbage, but you only get the wiring and install instruction if you buy THEIR overpriced garbage. Otherwise fuck you, plebian, figure it out yourself. Just give me a fucking PDF of the wiring you pre-dropped for fucks sake, I ALREADY gave you money, and you already spent the fucking time and wires...

The "owners manual" of a brand new camper I had to work on recently was more than useless. Practically anything you looked up didn't have any actual information, it was just a list of the 6 or so manufacturers that could be installed in YOUR particular camper - with the caveat that it could also be something not listed. And better hope they remembered to toss the paperwork into the messy drawer full of crap. Not that the paperwork actually told you anything, but...

Then the "inverted outlets" that were all labeled really nicely and literally say "inverter powered outlet" that they don't tell you is only actually inverted if you buy an inverter and install it. Not even in the owners manual. The manual literally says the outlets are inverted, but NONE of the damn things come from the factory with an inverter.

At least they could toss the damn manual in anything they built, it could probably cover every damn model since it didn't even remotely focus on one.

LLM's also aren't the only AI. They just so happen to be the most PUBLICLY KNOWN forms of current AI. And LLM's are specifically designed that they aren't going for direct intelligence. Some are building their own interesting internal "views" of what is is happening, but those are side effects of what they were designed to do.

There are tons of other trained Neural Networks that aren't directly public facing that are completely different. Like the medical imaging AI that is better than most doctors at reading X-rays.

And yes, they've found some issues with the training, sometimes from lack of data, sometimes from some unthought of other tells in the data. But who here can write a first draft so well that it can be submitted as a final draft with zero changes? No one, that's who.

What most people can't seem to wrap their heads around is the difference between AI, which just shows decision making intelligence, and AGI which will show GENERAL ( AKA "thinking" ) intelligence.

>But we want reports that are valuable, coherent and that actually tell us what's wrong so we csn fix it

You don't get that now 98% of the time. It's going to be difficult for the AI reports to be WORSE than almost all of the reports I've had to read through on the projects I follow.

It's just plain fact that the vast majority of bug reports suck ass, even from the users that bother to report shit back instead of just saying "it dun werk, this thing sucks" and use something else. If AI can help guide / ask questions about the issue / put everything together coherently, it can only help.

Jesus fucking christ...

Devs whine that users don't fill out bug reports with enough info. Some are downright hostile to the people ATTEMPTING to fill out reports properly, even who ask how to even get the info that would help the devs out.

Devs ALSO whine "we are anti-AI... for uh reasons" like the Thailand meme guy, even though it's designed to, oh I don't know, get the fucking information the fuckers need for a good bug report. Or at least present the information better than bubba's basic "It dun werk" filled out 50 times in the mandatory reporting template.

I mean god damn, this is one of the things that AI is a perfect fit for doing - taking the average moron, and making them intelligible in a useful way to people much smarter than them.

>for that you would have to claim that the LLM has personhood (and some other things).

If corporations have fucking personhood, so should the LLM owned by them.

>But those exceptions exist, and any speech carries the burden of respecting those restrictions. How does an LLM do that?

That is what the judge should have said in the ruling. Hmmmmm. Kinda absent in the sources I've seen so far. I'm not going to be bothered enough to read the actual court documents though.

>It doesn't matter, software does not have personhood. "Speech" as it was used, means "protected speech". LLM produce token sequences in response to input token sequences, that is not protected speech.

And you produce drivel in response to comments on every story on slashdot. Guess that addendum they have of all comments being the property of the commenter shouldn't apply eh?

"Maybe, possibly, could", there is also a chance that slashdot will post something that isn't a dupe and is also edited properly by the end of 2025. Doubtful, but it could happen.

It's the same old farts saying the same old shit as before.

"The typewriter is killing penmanship, letters and words now have no soul and feeling to them."

"Word processors are making people dumber, at least with a typewriter you had to think about what you wanted to say because you can't just delete whole sentences or paragraphs."

"Digital artwork is is killing REAL art like painting since you can just undo brush strokes."

"Digital photography is soulless and takes away from the REAL skill of setting every little thing on your camera and these youngster won't even know how to use a darkroom."

"Post processing digital camera pictures is cheating since you can push the image even further from what was really there than by using filters and specific films, or darkroom manipulations."

And that's just shit from the last 100-150 years. Knowing humans, Ug and Grog were probably bitching about those young caveman kids using that new fangled wheel thing, as they are now lazy and won't be as strong if they don't have to actually CARRY anything...

It's the last one. The stupid asshat put it full page behind the text, not just small in the corner next to the law firms signature area.

I fit wasn't full page behind the text I doubt the judge would have even noticed, much less said anything. The jackass had to put it in the most obnoxious place he could though, which makes reading the text more difficult.

>No, not at all. Reasoning models just generate each word as regular LLM models, then feed the answer into the same (or another mode)l to rewrite the response. One word at a time. There is no AI that thinks or reasons here, it's only really clever way of using the magic of "the next probable word". The only magic of LLM is "choose the next word". There is no thinking or reasoning at all, regardless of what they call the model.

You have no fucking clue what you are talking about. Period.

Here have an excerpt of the reasoning process for said isostacy:

But perhaps I'm overcomplicating this.
Let me try again:
Given that the object floats on top of another liquid (so maybe like oil floating on water), but here both have similar densities—0.97 and 0.99 kg/m, so very close to each other.
So in such a case, how much is submerged?
Wait if I use Archimedes' principle:
The weight of the object equals the weight displaced by the liquid it's floating on top of.
But wait—if you have two layers—like oil and water—the total buoyant force would be from both liquids. But perhaps since they are similar in density, maybe only a small part is submerged?
Wait but if I'm just considering one layer (the 0.99 kg/m liquid), then the object floats on top of it because its own density is lower.
So using Archimedes' principle:
( \rho_{object} V g = \rho_{liquid} V_{submerged} g )
Simplify: ( h / H = (\rho_{object})/(\rho_{liquid}) } ) assuming uniform cross-section
Therefore,
h=H*(\rho_{object}/( rho_liquid))
So plugging in the numbers:
( 6 m * (0.97 kg/m)/(0.99 kg/m ) = h )
Calculating that: approximately ( 6*0.98 5.88 meters submerged.

But yeah. It's just estimating the next word. Sure. Oh wait, that wasn't the ANSWER, that was a snippet of the dump of the reasoning chain that went on in the background. Not something that "normal" users would ever see.

>An undergraduate would not do that, it would not select a wrong word and then completely miss the mark because the selected word is more associated with flowers than electrical engineering.

I've had undergrads who couldn't guess the correct amount of days in a year. On a take home lab sheet. That they had a week to work on, and full internet access for. There are a significant number that of them also that can't do simple logic problems that a 3rd to 4th grader should be able to figure out.

So I call bullshit on your magical full classrooms of smart undergrads.

>It doesn't reason.

Except that that's exactly what the latest "reasoning" models do. You can even peek under the hood and see the reasoning steps it goes through if you run it locally.

Especially with the local versions of deepseek - ask it an isostacy question, give it some densities for some parts of the equation, and it reasons in the background that it needs to use the Archimedes Principle and will fill in the rest of the densities that you are asking. It's actually quite interesting watching it do the reasoning for each part of the problem.

>It doesn't understand.

It "understands" about as much as a typical undergrad. It can do many problems without having to be told the exact steps needed since the underlying principles are well known and in the training data. It doesn't know specifics, but it can use logic to put together the steps needed to get specifics. At least for math. Obscure History or esoteric pop culture... not so much.

You've obviously forgotten about the usage surveys Photoshop used to give out then. That's also telemetry, just much slower, and voluntary. They even used to offer discounts for the surveys as incentives.

Too many people just plain didn't bother.