Posts by Groo The Wanderer - A Canuck

NASA's Artemis III will need three rockets to do the job Apollo did with one

It's that "not blow up" that they seem to have trouble with...

AI ops tools will create console sprawl and break IT more often: Gartner

Artificial Ignorance breaks everything more often, not just infrastructure.

Top EU court clips YouTube's intermediary defense over reviewed content

I can see why "I reviewed it before signing the contract" negates any claim that "I didn't know what they were doing."

Europe's chip ambitions won't break dependence on US cloud and software, says Forrester

Why do I have a feeling Forrester is an American company firmly in the throes of self-denial about the future...

SpaceX Starship Flight Test 13 takes issue with the 'flight' bit

Space-X: The overvalued government welfare recipient that fooled the world into thinking they were capable of delivering on NASA's needs...

Chinese memory ban would cut off RAMpocalypse relief

That would keep American memory prices high, but not matter one whit to the rest of the world if China ramps up production. I mean how do you seriously think you can "infect" a memory chip?!?!?

Methinks the American politicians' brains are infected with some brain-eating parasite...

At last, a good reason to buy an AI PC: Reining in runaway token bills

Re: You don't need an "AI PC" for that

llama.cpp is your friend if you lack VRAM.

Rather, I look forward to the day when the industry stops trying to bolt "AI" on everything whether it makes sense to have it or not, and gets back to work solving business and industry issues instead of lining the already-far-too-deep pockets of Altman, Musk, et. al.

In the meantime, I'm expecting the announcement that Microsquishy has released "Copilot Calculator" any day now...

AI vendors have found someone to pay their infrastructure bills: You

Re: Really...

I like that one! :)

Everybody pays for the AI bubble and it's bursting in the near future except the tech-bros at the top who are running the scam...

They're very insistent on "buyer beware" type clauses for using their "products"; they guarantee nothing, and promise nothing - it is all on you to double check the "intelligence" outputs.

OpenAI admits GPT-5.6 occasionally deletes files – but it's an 'honest mistake'

"Ooops! I deleted the production configuration files again! My bad. Sorry 'bout that, but you're supposed to review everything and anything I do before you hit the Ok button. It's your fault I screwed up."

- AIBot

TSMC's $265B US fab pledge is the outline of a concept of a plan

But you can count on the politicos making a big deal about it and claiming credit for something they had nothing to do with in reality; 'tis the nature of the beasts.

Linus Torvalds tells AI haters to fork off

I don't hate "AI". It just think that LLM technology in the field of Artificial Intelligence is a red herring and a long-term dead end for many of the things they're still too naive to realize it is very ill suited for. Why do I say that? Intuition. I trust my own intuition about the limitations of statistical event processing and handling general case real world situations.

I do not believe for an instant that it is possible to make LLM technology predictable enough to be safe enough to use for fields like self-driving vehicles, for example, much less relying on it for health care! I expect all the current-generation would-be providers to lose their deployment authorizations from the licensing bureaus around the world soon enough.

Microsoft cancels Patch Tuesday for some Dell users over surprise shutdowns, overheating devices

GenAI "patches" - the gift that keeps on giving.

Patchpocalypse Now: Microsoft tops last month's record with 622 Patch Tuesday CVEs

The workload they're dumping on the administrators of the world who have to review those CVE's before they're allowed to affect corporate production is quite insane...

Musk promises purge after Grok Build caught sending entire repos to the cloud

This kind of thing is always going to happen as long as you're relying on statistical prediction for "logic processing" of your "intelligence" aspects. Statistics are randomly unreliable and incorrect, triggering wrong events, tokens, and responses in response to their inputs. It really is quite unavoidable, because the mere fact that you are doing a statistical aggregation to form your LLM's driver data set means that you are losing details in the process, and introducing inaccuracies.

Sticker shock has execs rethinking this whole AI thing

If you think it is expensive now, just you wait - the AI vendors and providers have just started raising their prices to the level where they're not losing money on the services, but they've yet to raise them to the point of being profitable yet. How the hell a non-profitable company has a near-trillion dollar valuation is beyond me, but that's the US stock market for you...

Zuck's AI ambitions put Meta on course to become America's next big cloud provider

"Both are immensely profitable, earning $132.2 billion and $60.5 billion in profits last year, respectively. That's profit, not revenue."

Which is why we're stuck with AI generated spam forever. The spammers pay Meta, Google, and everyone else for the spamvertising ads they spew to fleece people with.

The price is wrong: AI cost calculation has to consider task completion rates, not just token costs

Let me know when there is an LLM capable of handling a 5,000,000+ line system spread across a dozen modules... they're single module toys at best.

Memory makers are slaves to the boom-bust rollercoaster, and the AI boom is the wildest ride of all

It certainly is a big bubble, and I can't blame the vendors for wanting to cash in, but leaving the desktop and laptop markets in the lurch is disgustingly greedy and despicable.

KDE Plasma users face a dire omen of change: 6.6.6 arrives

I'm not citing SHIT to prove SQUAT to you when I'm talking about real-world personal experience with Debian 13!

The problem is most Wayland drivers render the initial image to a software buffer and then push the whole thing to the display. That slows down rendering performance dramatically compared to having the card do direct rendering as OpenGL does with X11 displays.

Any desktop which "mandates" Wayland instead of X-11 for the sake of my gaming performance and CUDA stacks is welcome to take a leap off the cliff of complete and utter irrelevance due to the absolutely abysmal performance and support Wayland has for my game catalogue. Screw "waiting for the next release"; x.org works now!

Meta admits its first ‘superintelligence’ was too stupid to survive for three days

Damned users, always breaking perfectly good code. *LOL*

AI slop writing has taken over the internet, particularly LinkedIn and X

That's because the users of those services are largely uneducated borderline illiterate fools and suckers who are absolutely desperate to appear "smarter" than they are. Not to mention all the AI-generated spam "articles" that get posted to try to get you to follow clickbait links to advertising-laden nightmare sites. (You know the ones I mean - the ones that show you two paragraphs of an article, 15 ads, and a "next page" link to two more paragraphs with another dozen-plus ads.)

EU 'Chat Control' snoopfest returns after vote to kill it falls short

I'm very surprised the EU would allow unfettered, unlimited scanning of citizens personal accounts without a warrant; that flies in the face of every piece of privacy legislation they have ever enacted. I think it is clear that any such scanning should require a warrant from the courts in any nation in the civilized world. Uncivilized hell holes are likely to opt out of the limitations and spy on their own citizens without limit, such as the way China monitors their internet for "dissidents."

Court tosses Microsoft's appeal in pre-owned software licenses spat

Nothing like a juicy and pleasing tidbit of news from the UK courts to start the day off with!

Boffins bet on quantum computers, AI supers to solve fusion fuel dilemma

"...and then a miracle happens, and boom! Profit!"

Zombie ‘who owns Unix?’ lawsuit comes alive again

And in a desperate hail-mary attempt at extorting money from a reputable corporation, yet another pathetic patent troll...

Confidential computing's trust mechanism is broken. The fix may not exist

And once again marketing bafflegab trips over the rock of hard reality...

Startup targets datacenters with 3D-printed nuclear reactor module

Interesting. It sounds like an eminently practical solution to add to humanity's creative workarounds for climate change and energy source stability.

AI bills are baffling the C-suite after shift to usage-based pricing

The costs are easy to understand. The requirements are simple: You bend over and scream, and the Artificial Ignorance service providers rape the ever-living shit out of you and your wallet.

Windows 11 can now turn back the clock when updates go bad

Is that all? It seemed like an eternity of suffering the slings and arrows of unwanted Artificial Ignorance before I took my final leap into the welcoming arms of Debian 13 with a Mate desktop (viva la Gnome 2!)

Sure I believe them. After all, they only bragged about their snapshots 10-12 years ago when Windows 11 was first released.

What I hear them saying is that either they lied about the capability then, or they're lying about it now, but they can't claim credit for doing the same thing twice, even if they did screw up their first attempt at it.

How the AI bubble could pop and take down the global economy, according to the BIS

The bankers and investors are first in line with claims on any actual physical assets these "businesses" own. The rest is just digits or paper and not worth one whit without the investment community's continued confidence in their stock. Once confidence starts getting shattered by the inevitably high number of failures to improve anything in the end get reported (what are the current numbers? 60+% reporting no or negative ROI?), and the number of GenAI projects and deployments being cancelled in favour of returning to human capital investments (e.g. Ford engineering) starts to get noticed, the whole bubble comes crashing down and the bankers and investors take over the physical assets and everyone else is left scratching their gonads...

Really? You just finally noticed something that the sane members of The Reg have been commenting about for a couple of years now? After a few more hundred billion has been put to risk by the hype-investment scams?

Changing AI math could reduce the hardware burden, researchers show

Interesting idea; every bit as interesting as the recent paper about one-bit encodings with separate valuations indicated by the on-state rather than fixed values.

It still doesn't make LLMs actually intelligent, but it does make them more palatable for the initial mapping of spoken/written text into semantic symbols for further processing.

Boffins peg narcissistic leadership as the real driver behind 'return to office' demands

Bingo. This in every denomination you care to spend.

There are no shortage of '50s and '60s mentality managers in play even in the most modern of companies, who flat out don't trust what they cannot see and berate in person, off the record, away from video conference cameras and recording capabilities...

AI may be good at finding security vulnerabilities, but it can't beat human stupidity

Never underestimate the sheer power of the foolish mistake, the fly in the ointment of data processing...

AI giants back non-profit to retrain workers left behind by AI

Re: "Corporate incentives to retrain and redeploy workers"

Disagree completely. Absolutely decades of B-movies have proven through sheer documentary volume that your minion robots will become intelligent and rise up against you in your volcano lair.

You just can't argue with the standardized plot of the B-movie; the villain always gets their comeuppance! :D

And the suicide of society for the benefit of techmoguls and billionaires continues unabated.

I guarantee you won't make as much when "retrained" as you did before. Not a hope in hell of that.

Never mind the fact LLMs being treated as "intelligent" is complete and utter fraud, some other technology improvement will eventually deliver what LLMs are being advertised as capable of doing, where the LLMs actually fall down. As far as I'm concerned, the only valid use for an LLM is the initial parsing transformation of inputs into back-end tokens for processing by specialized code, not for mere statistically driven regurgitation of predicted outputs instead of the accurate outputs the theoretical specialized code provides. They are not suitable for the actual intelligence processing, world view and memory models, or transformations of the models into outputs.

Trump-shuttered climate change site back online in nonprofit hands

Fortunately the world is far more sane than the Orange Menace also known as Der Pumpkin Fuhrer, whose idea of "science" is a conspiracy theory post somewhere on Bullshit Social...

IBM stacks up a sub-nanometer chip future

Re: IBM Research - why ?

Don't underestimate being the primary supplier of compute resources for the financial and insurance industries alone; those have a huge impact on society. They may not have the market share they used to, but they've held on to their key clients that give them some clout in the market.

A finally started shipping technology that has virtually no traction in the market yet and won't for a couple more years, at best, because chips need redesign to move to the smaller processes. By the time those products reach full volume, I guarantee IBM and TSMC's "research" will be entering production, which is what I've claimed from the start. Not that Intel isn't finally starting production after over ten years of talking it up.

Just proving my point of a couple months ago that Intel was behind the times with their "18A" and that by the time they hit full volume, others will have left them behind. Intel has been behind on the development curve for nearly 20 years now, living off reputation no longer earned.

Micron locks in historically high memory prices for five years

Re: Is This A Prediction?

Really? The bubble has started quivering? Then the burst shouldn't be much too far down the road... 'cause the whole damned thing is nothing but a pyramid scam for inadequate and over-hyped technology.

Re: Silver lining?

True, but the simple fact of the matter is that graphics intensive games and user interfaces of the modern age require huge amounts of memory compared to the green screen era you're referring to, when software did only what it needed to for the specific job it was designed for, instead of importing the Universe knows how many graphics imports, transformations, and video embedding capabilities that are considered de rigeur for modern applications.

You know, when the pyramid scam and stock market bubble of Artificial Ignorance bursts, the market is going to be flooded with RAM and those same companies are going to deeply regret these contracts.

I predict mass bankruptcies across the US and around the world when it happens.

Artificial Ignorance is just statistical prediction based on mass data scraping. It is not "intelligent", has no concept of "meaning", a "world view" model, a long term memory, or the ability to even keep track of more than one chat session at a time and learns not a single thing from prior sessions.