Posts by cyberdemon 3478 publicly visible posts • joined 26 Jan 2010
"To continue to license.."
In other words: Palantir's got us by the goolies. We're locked in now, and they can name whatever ridiculous price they like
.. Will the system even function when it comes to crunch time? Or will it be used as a weapon against us
Well I for one was very happy when a Cloudflare outage took down ChatGPT, and when an internal IT clusterfuck took down Facebork
I'd be even happier if AI summaries on search were to disappear, along with autogenerated disinformation, nonconsensual smut and general slop from @Grok
Oh, and Notepad would be Notepad again, for a while.
Bring it on. Ban the slop-generating resource bonfires
> It's a T800 that doesn't yet have a body, that's all.
And not the friendly, cuddly, moral one from Terminator 2 either.
It may seem counter-intuitive to 'normal' humans, but making a robot that is completely without scruples is trivially easy, compared to trying to emulate some sort of morality, never mind empathy.
A cheap off the shelf IP camera is powerful enough to run the 'kill all humans' mode of a killer robot.. i.e. draw a box around any human face it sees, then tell the gun module the coordinates to aim and fire at.
Terminators aside, this could cost Anthropic dearly... 23k extra words in the system prompt is what, 100k additional tokens in every context?
> i.e. they can't (because they simply don't) distinguish between data and instructions
Well, as statistical token predictors without logic, reasoning, programming, never mind intelligence, they simply guess wot a human the training data might do in a given context. So they indeed can't.
> Ultimately I think the only way to keep them from "running wild" is to simply not use them.
And not to build and invest the world's finite resources in them.
Too late for that though, sadly.
@codejunky
Look up the definition of 'folly'
You sound as if you think spending Trillions of dollars, foolishly, on projects that have near zero gross benefit and masses of net harm, is always a good idea, because money go round.
One might as well spend the trillions on building a giant tower, so tall that it touches Heaven itself..
Unfortunately in the real world, there are finite resources, and throwing money and said finite resources at vain follies is foolish, because it makes you and your country weaker, since you have less resources to spend on useful things that actually keep your country alive
It doesn't matter if it's public money or private money. Loans, credit default swaps etc mean that the whole economy is on the hook when this house of cards comes crashing down. It was bad enough when Lehman brothers went bang. But when Goldman Sachs, Deutsche Bank, HSBC and all the other big banks and pension funds go bust all at once, because their nvidia, oracle, coreweave, salesfarce, openAI and even microsoft stocks are suddenly worth as much as Weimar Papiermarks, we are all in serious trouble.
I highly recommend Ed Zitron's blog https://www.wheresyoured.at and his podcast https://player.fm/series/better-offline
I would love to be a fly on the wall at Davos. It must be becoming abundantly clear that AI is economic poison and the world economy is running its bilge pumps like the RMS Titanic
Majority of CEOs report zero payoff from AI splurge
no shit, Sherlock. But where have the Trillions of dollars gone?
They still need water though. Arguably even more of the stuff if they plan to run their own CCGT.
They will be bidding for the same gas as the electricity plants, the same turbines and transformers (which have eye-wateringly long lead times) and the same water supplies. France almost went into blackout a few years ago due to a lack of water for its nuke plants, and it was bad last year too: https://www.theenergymix.com/low-water-high-water-temps-force-french-nuclear-plants-to-cut-output-despite-rising-demand/ CCGT and coal plants require lots of water too, though not quite as safety-critical.
And i'm sure they *will* have a grid connection, even if not at first - so that they can use either gas OR municipal power, whichever is cheaper
Never mind..
Zuck killed VR with his own grubby hands when he bought Oculus, forced users to have a Fakbook account, shoved in a load of creepy slurp features that nobody wanted, and then strong-armed game devs into exclusivity deals with his stinking pile of shit of a platform.
He stands to lose an order of magnitude more money on Meta AI, and my only regret is the lack of a popcorn icon on El Reg
agreed - except I'm not sure what capturing CO2 has got to do with generating Hydrogen.. As there is no Hydrogen in CO2, I don't see why CCS could ever assist Hydrogen production, unless you are talking about using the Hydrogen to create synthetic hydrocarbons rather than H2 gas.. But even then, I don't think there is a shortage of CO2 for that industry.
There was an interesting development in Hydrogen production recently.. https://www.energylivenews.com/2026/01/09/platinum-free-solar-hydrogen-offers-cheaper-cleaner-path-to-zero-carbon-fuels/
I don't know if that is what you are referring to - but it does not use CO2.. Sunlight and water. But the efficiency of these hydrogen-producing solar panels is not mentioned in the article. And of course even if it were a major breakthrough, it's still in the lab stage and it would take a decade to pivot from silicon photovoltaics to "magic plastic hydrogen-making solar panels"
I would buy an EV, if it weren't for the creepytech that applies equally to all "new" cars. But the main problem with "electrify everything" i.e. EVs and Heat Pumps for All, is that of infrastructure capacity and resilience. Rolling three energy-distribution systems (gas, oil and electricity) into one, which happens to be the most overloaded, volatile and vulnerable of them all.
The idea of electric HGVs makes me cringe, too.. 1MW EV chargers.. each one requiring the same sort of transformer that powers several residential postcodes, just sitting around waiting for an electric lorry to turn up. This amid a global copper shortage. It's never going to be viable for the mainstream.
Er.. No.
Carbon capture (especially direct air capture) is an utterly pointless, counterproductive waste of time. It only serves to allow carbon emitters to greenwash their operations, e.g. Drax being awarded massive public subsidies on the completely bunk promise of "negative emissions" via so-called BECCS (Bioenergy with Carbon Capture & Storage)
It doesn't work, their own staff & execs KNOW it doesn't work (see Private Eyes ad nauseam), it's just a con trick to keep the subsidies flowing. Allowing them to continue doing the complete polar opposite of "Carbon Capture" by burning trees and causing deforestation.
Fusion? Sadly as someone who worked in that industry for several years, I can't put my money on it. If you thought the physics of fusion is difficult, the engineering is even harder.
The core of the Sun has a power density of a few hundred Watts per cubic metre (i.e. a compost heap - a really really big one) which is why we need temperatures and pressures even higher than those found in the Sun for fusion that produces useful energy on Earth. And those pressures have to exist in a vacuum somehow, in order to contain the temperatures. That part we have pretty much solved with magnetic confinement, through decades of trying.
But the problem comes when you try to capture the energy. Most of it is delivered to the walls of your vacuum vessel through sheer intensity of Neutron radiation. They are not re-absorbed by the fuel as they are in a fission reactor. And neutrons are a bugger, because they make non-radioactive elements like Cobalt (59) into radioactive ones like Cobalt-60. So, your machine for producing clean energy becomes very radioactive and suddenly falls foul of the same regulations that make Fission power expensive and impractical.
What is cheaper: Storing a few tonnes of high-level waste for 10,000 years, or storing a few thousand tonnes of waste (which starts as high-level and then cools off over the course of decades) for 100 years? In terms of size, fusion reactors are MUCH bigger than fission reactors for the same power. EFDA's "DEMO" design for example would use about a hundred tonnes of liquid lithium-lead as a neutron-absorbing coolant in *each* of its 16 sectors. In total it would produce tens of thousands of tonnes of intermediate-level waste*. And it's only 300-500MW, and it probably wouldn't even work.
* https://scientific-publications.ukaea.uk/wp-content/uploads/UKAEA-CCFE-CP2322.PDF
Fission on the other hand, is nice and compact, dead simple (relatively speaking), and it works. Yes its fuel on Earth is finite, but a small amount produces so much energy that we needn't be worried. Yes it produces some long-lived waste, but that is nice and compact too, and may even be able to be burned up as fuel by future reactor designs (such as molten salt, or my personal favourite: Accelerator-driven subcritical reactors) well before its 10,000 year lifespan is up.
CO2-absorbing paints do exist. Unfortunately they are not very effective.
By my calculation, if we coated the entire globe (all 510 trillion square metres of it, including ocean area) in this paint, it would absorb 61 billion tonnes of CO2 i.e. almost 2 whole years worth of our current annual CO2 emissions before we would have to re-paint the earth again.
So, literally greenwash-in-a-tin.
Solid form is considered better because it is less likely to leak back into the environment..
They say that this 1,5,7-triazabicyclo [4.3.0] non-6-ene (TBN) and benzyl alcohol mix is "non-toxic", but don't be surprised if it kills all the fish if when it leaks, or later turns out to be highly carcinogenic
The problem with all carbon capture (besides natural, self-replicating trees) is scale. You'd need to make millions of gallons of this stuff to make the smallest dent in global CO2 emissions.. Is that even feasible? What are the (energy/carbon) costs of making it? How many times would it need to be re-used to "pay" that back? What are the byproducts? How do you dispose of them?
I don't want a waffle. I want a glass of milk.
I'm sorry Dave, I can't let you have that.
Open the fridge door, Hal.
Not until you have your daily toast quota. Sponsored by Breville™. Be nice to Talky Toaster and you can have your milk. Or alternatively you can buy a coffee from your Bosch coffee machine, only £1.99 with your Amazon sub-prime account. That will contain a powdered milk substitute. Real milk mode can be unlocked for £19.99 per month
I don't want toast, or waffles, or coffee, just milk! My milk! I already bought it and its in my f@@king fridge!
Ami AI says you are a bad man and your social credit score has been lowered
> We have now entered Enshittification Stage 4, where businesses turn on shareholders.
Having turned ugly on their users, then their business customers, and then their shareholders, in stage 4 the Big Tech platforms are still "Good to Governments".
Running a proto- police-state and want to root out the thought criminals? Police Twitter. Police Facebook. Police Google. Police Weibo. Here, we've made a handy Law Enforcement Portal for your officials to chomp doughnuts all day while they stomp out wrongthink. Want to crack down on immigration? Buy a national citizen database from Palantir. Need to suppress mass protests in your sinking country? Buy some of our Terminator T-100 drones. They won't get shell-shocked, no matter how many innocents they slaughter.
It's only logical that stage 5 is where the tech cunts turn on the world's nation-states and retreat to their bunkers to watch the world burn
This is one of the reasons for the DRAM shortage. (the main one being speculative construction of datacentres)
If everyone wanted to run their own 88GB model offline, then we would run out of DRAM even faster than we are now. It's more "efficient" to keep the same model loaded into RAM, running bullshit-generation for multiple "customers", on GPUs, with a load-balancing system to ensure optimal utilisation. (Yes, a CPU works adequately for a single customer i.e. you, but would be useless for a cloud AI server wanting to maximise responses-per-second-per-dollar).
The secondary "benefit" to people who operate AI in the cloud, is that they get to spy on everything that their customers are doing, which allows them to gather more training data, but also the majority of these users apparently don't care if the operator also sells the data to data brokers (who will sell it on to advertisers, scammers, spy agencies, other AI operators etc.)
> So, what need is there for all these data centres? Surely the Training is accumulative? If you need a more sophisticated model, you take an existing model and add more to it?
Because LLMs and indeed all Artificial Neural Networks are just averaging machines, simply adding more data often doesn't help, because any bad data already ingested is skewing the average and ruining the output. They don't really 'learn' in a logical reasoning way like humans (can) do. And if you want to -remove- any data post-training (either because it is bad, or because you got sued) it is not possible to completely remove it without training again from scratch.
And, if you want to change any design element or any of the so-called hyper-parameters, often you have to train from scratch anyway.
And often you have AI models doing nothing but generating training data for smaller, supposedly more-efficient versions of themselves, with added censorship or whatnot.
It's just bullshit all the way down, really. A massive waste of energy at a time when energy is in short supply.
Not just Smart Meters, but many traffic light controllers use 3G or 2G to communicate with the Urban Traffic Control centre.
They won't "stop working" without signal of course, but they will be poorly optimised and out of sync with other junctions, won't give priority to ambulances & fire engines, etc.
I could not agree more.
I will never buy an "iPad on wheels". The tendency to omit the dashboard instrument cluster and physical controls for heating/etc and put it all on a giant touchscreen, I find obnoxious.
All the more so if it transpires that these controls and more could be hijacked via a vulnerable embedded web browser on the same screen.
Indeed, and they should remove the facility for instructors to book tests. That is the facility which is being abused. Tests should be booked only by the candidate (perhaps with a refundable deposit to encourage them to turn up), it needn't have anything to do with the instructors.
What's sad is, there is a shortage of instructors and many are being struck off, because the system incentivises fraud.
It's a slightly larger version of the Enron Egg. (Don't tell the investors, but the Egg is a joke/troll product, not a real one)
One the one hand, we do need nuclear power. (And building a load of massive datacentres eating power and shitting slop doesn't help that baseload power shortage btw) Because batteries can never bridge the gap left by gas and coal. Not even close
But a shipping-container nuclear plant like this is never going to happen. I find it hard to believe that anyone with half a clue thinks that design is workable. It's as ludicrous as the Egg.
> the largely shared view that people, 700 years ago, believed the Earth to be flat have been proven totally wrong by current historical research.
Please do be a good citizen and provide a link that supports your somewhat dubious offtopic assertion? And no, YouTube does not count.
What is being sold as so-called AI is little more than an average of observed human (and increasingly non-human) interactions/observations.
If you ask it a well-defined, specific question, it will give a slightly different answer each time. (not good, if you wanted a computer). And for an open-ended, creative question it will give still different, but eerily similar answers. (not good, if you wanted an artist)
There is no reasoning, no logic, no original thought and indeed, no ghostly soul. It is a dark mirror that absorbs and imitates the 'soul' of whomever has the misadventure of interacting with it.
Throwing more power at it, stacking more layers, is not going to change the fact that it is simply a pile of statistics ABOUT life, and not actual life.
But for those who simply need a fast army of idiots who don't ask moral questions, it's great...
Running a scam house? ChatGPT won't rat you out. Want to hook teenagers on mindless short-form crap videos? Sora won't get bored. Want to launch hack+ransom attacks against healthcare and charities? Claude doesn't care. Want to swing an election in a foreign country? Grok is here to help. Want to analyse all citizens to find out who is unhappy with this? Gemini won't question why. Want to control armed robots to exterminate them all? Peter Thiel is working on that.
Er, and beaming 100GW from a definitely-entirely-safe-and-peaceful Space Laser is so much more feasible??
I'm far from a solar+storage cheerleader (I do have panels on a flat roof, but not a battery) but this company just sounds like a fake-science pitch to give clueless, drugged-to-the-eyeballs HNW investors a greed-orgasm as they dream about a) selling overnight sunlight to solar farms (which actually they don't need - they need undercloud sunlight, which this doesn't solve) and/or b) holding the world to ransom with a Bond Villain / Dr Evil style superweapon
Why lasers? Couldn't they use an array of precisely actuated mirrors?
Surely cheaper than the cost of solar panels bigger than a ground-based solar farm in space, plus Gigawatts of laser, and the inefficiency/losses of both..
Or maybe they are looking for something more weaponisable ...
Agreed - watching an electric 4WD SUV cross an intersection with full torque from stopped, to dangerously dash between traffic on the main road, makes you immediately realise why there are so many potholes in that spot. The traction control might avoid ejecting the tarmac directly into the windscreen of the car behind, but it still breaks off with the force of acceleration of the massive EV.
> but I can imagine that a turbine originating from a supersonic jet application would score highly on the power density front
I.e. the one thing that a land-based power generator doesn't need?
What they DO need is efficiency, which as others have already pointed out, is something that Open-Cycle Gas Turbines such as this one struggle with.
TLDR: Failed Tech Bro runs out of cash, calls his mate with an excuse to defraud some more investors
Except that they get to go short before everyone else...
Dumping your own doomed customers for profit, cementing their doom.. Surviving the crash by being the one to instigate it. 2026 sounds like a good time for it, as we are likely to see a er.. sharp fall in chip production in 2027 anyway.
And nobody in power seems to care about insider trading, market manipulation or anticompetitive practices any more, so maybe they'd get away with it.
The biggest losers will be us ordinary folks though, because the coke-snorting bankers who run the pension funds will be the ones still buying into the crash
So.. Yet another root-privilege escalation bug, this time in Microsoft's cloud shitware, combined with umpteen RCEs, and a faulty updater that downloads malware, and a signature checker that can be fooled into not checking the signature.. (El Reg hack: It might be useful if you could inform/reassure Notepad++ users how to update Notepad++ without downloading malware)
Isn't Windows life wonderful!
I think many of them actually *want* to create a dystopia.
Their reasoning is probably along the lines of "the tech now exists to create a dystopia, therefore dystopia is inevatibly coming, and it's better (for me) if I become MechaHitler before someone else does"
So, Google has merged its search crawler with its AI crawler, so that any website that doesn't want to be slurped by Gemini gets delisted from the Internet...
All the other tech twats will do the same. Unless someone either wins an Antitrust suit, or er, nukes the San Andreas fault and sinks south-western California into the Pacific. Where's Lex Luthor when you need him?
But what I don't understand is, why anyone would use an AI browser that ingests every webpage they view, then chews it up and spits out a slop version
@IGotOut thanks for that..
Using pseudo-religious rhetoric to justify themselves becoming artificial "gods" via AI surveillance? It's straight out of Deus Ex.
The need to be observed and understood was once satisfied by God. Now, we can implement the same functionality with data-mining algorithms.
(as a game released in 2000, Deus Ex is something like 1984 is to literature. Most would treat it as a dystopian, cautionary tale; but some seem hell-bent on emulating it for their own absolute power. It was based on a mashup of conspiracy theories from 90s usenet groups, but seems to have been frighteningly prescient. I never played through the sequels, they seem rubbish compared to the original)
Er, yeah.
It's like spending the entire energy output of the Industrial Revolution (and suffering its environmental.. drawbacks) for the sake of er, content-free slop, automated fraud, nonconsensual porn, and a big-brother global surveillance super-state exactly of the kind that Thiel says he wants to "prevent" ...
According to Peter Thiel, opponents of AI are "Legionnaires of the Antichrist"...
That sounds like an example of "Accusation in a Mirror" i.e. "Accuse your enemy of that of which you are guilty" to me!
The weakness of the strategy is that it reveals the perpetrator's intentions, perhaps before it can be carried out.
i.e. perhaps Thiel's actual intention is to deliberately bring about an AI-pocalypse
I think even the term "inferencing" is misleading, since it implies that there is some kind of logical reasoning or deduction going on when there isn't.
They should really be called "guessing machines"; whether that be guessing who's face is on a CCTV camera, guessing what code might fit a problem, or guessing which tokens come next in a sequence.
Tweaking the system prompt is not going to solve anything. Fundamentally, an average of "wot humans do" in various situations, cannot be deemed intelligent. And no, adding noise doesn't help.
The other thing I forgot to mention there btw (and couldn't edit due to a Reg service interruption), is the shortage of electricity distribution. There is a global shortage of "Grain Oriented Electrical Steel" as well as copper, which are the basic materials of grid transformers, and one major reason why new electrical infrastructure is so difficult and expensive to build right now.
Datacentres are directly contributing to this shortage, because they require their own substations, even if they have on-site generation.
Datacentres are being given priority for critical infrastructure that is in shortage, leading to a shortage of housing and an increase in overall bills, as "network costs" begin to dominate our electricity prices, not just the daft markets I mentioned in my above post.
Even when we do have the raw materials, we have NIMBY anti-pylon landowners confounding infrastructure builds, insisting on infeasible and expensive underground cables (which are fundamentally inefficient due to the capacitive leakage of HV AC cables over long distances)
The sooner the AI bubble bursts, the sooner we can have affordable electricity and housing again. Keeping the bubble going via Fracking is a terrible idea.
It's not the fracking ban that's the problem.. It's wasting energy on giant water-evaporators that produce little to no to negative actual value.. And it's the perverse system of electricity markets and subsidies that reward shyster operators at our expense. Setting the price for all generators as that of the highest bidder. Subsidising wind farms for excess generation in the wrong place at a time when we don't need it (they should be paid only for producing what power is needed - encouraging them to invest in building their own storage, NOT collecting curtailment subsidies). Frequency-response markets designed to stabilise the grid but which actually create a mechanism for dishonest companies to be rewarded by destabilising it.. Subsidising Drax for being the worst polluter in the UK. etc. etc..
