Er, are you using the slow Fourier transform? You should try the fast one
Posts by cyberdemon
3170 publicly visible posts • joined 26 Jan 2010
Page:
So much for watermarks: UnMarker tool nukes AI provenance tags
UK VPN demand soars after debut of Online Safety Act
Maybe the idea is to get everyone onto VPNs, because it makes surveillance easier. All the naughty people in one (or few) places, send a few subpoenas and superinjunctions, backdoor some VPN apps.
Much easier than legal battles with ISPs i'd guess, and politically easier with the "think of the children" arguments
Windows 11 is a minefield of micro-aggressions in the shipping lane of progress
First release candidate of systemd 258 is here
White House bans 'woke' AI, but LLMs don't know the truth
Britain's AI datacenter plans face energy, planning, investment challenges
Re: The cost of energy in the UK
Not so!
If you read this week's Private Eye (p38) apparently Teesworks Ltd (the dodgy freeport run by Chris Musgrave and Martin Corney) bought a "private wire" electrical distribution network, including 20 substations, for er, £10.91 - 55p per substation - from publicly-owned South Tees Development Corporation, thanks to their mate Lord (Ben) Houchen
They now plan to locate datacentres on the site. And because it is a freeport, it would be exempt from various taxes and possibly data protection regulations.
So it just goes to show that if you have the right (political) connections then you can get all the (electrical) connections you need, paying only pennies for infrastructure worth tens of millions of pounds. It's corruption to the core of government. Another stinking Tory legacy that so-called Labour are happy to continue..
Sh!t happens, so Microsoft is paying biz to flush its carbon sins underground
If you're forced to use Windows 11, here's how to steal some of your time back
Re: Grief - that looks horrible
Just what I thought
I'm on Debian for the last 20 years, the closest I go to Windows is my Dad's computer which is stubbornly still running Win10.
My first thought on seeing those screenshots was the same as yours: Good Grief, is it THAT bad now? You can't see for trashy clickbait "news" headlines and manipulative psyop shit. It looks like your PC is running in an iframe inside Facebook, or some expired-domain-parking site
Forget the popcorn, Where's the sickbag icon?
US signals intention to rethink job H-1B lottery
Curl creator mulls nixing bug bounty awards to stop AI slop
> Maybe use an AI trained to spot AI submissions?
That would be like trying to use a blunderbuss to shoot down incoming bits of shrapnel from a blunderbuss
Not going to work.
AI is pretty useless at debugging in general.. One of my CompSci profs gave us this adage: "Debugging your code is at least twice as hard as writing it - therefore if you write the cleverest possible code that you can, then you are, by definition, unable to debug it"
For "AI assisted" debugging it's worse, because it will lead you down all kinds of rabbit-holes and wild-goose chases, wasting your time, its energy, and ultimately nobody gets any smarter for it.
xAI's Grok lurches into right-wing insanity, offers tips on assaulting man
"The response should not shy away from making claims which are politically incorrect"
So now it matches "politically incorrect" in its prompt with 4Chan's /pol "politically incorrect" channel and spews bile dredged directly from the arsehole of the Internet
This is not "programming", it's simply garbage. It's time we cut off the electricity supply for these noise machines
AI coding tools make developers slower but they think they're faster, study finds
Automated Ignorance < Human Genius. Story at at 11
Really, is it surprising that statistical bullshit machines are incapable of "tasks" requiring actual thought and analysis, when they are simply (polluted) statistical models of wot (supposed) humans have said in previous scenarios?
It is umbilicism on a civilisational scale to believe that a mere model of our past behaviour can now surpass ourselves to augment or replace human intelligence, based entirely on what has gone before.. And without any individualistic spark, curiosity, or hope
I worry not about superintelligent AI but apocalyptic mediocrity i.e. we are replacing all of ourselves with a simulacrum which behaves (statistically, without any actual empathy, logic or indeed intelligence) roughly as a thoroughly mediocre 2010s human might, but at a scale and speed limited only by how much energy we can burn.. Yet without any capacity to understand, improve, invent..
Constantly we feed our conscious 'souls' into this infernal contraption, in the vain hope that it might give us some temporary boost in mental output, and drive profit/growth before we all get bored of it.. Meanwhile the malign uses of such a simulacrum (dare I say homunculus) grow all the more powerful. Misinformation, disinformation, scams, fraud, attacking the weak without empathy or trauma (a feature, not a bug), building automated tools of oppression (e.g. autonomous lethal drones, which ate now terrifyingly real), and mass surveillance and control on a scale that the Stasi could nary have dreamed of) thus facilitating the destruction of our current civilisation.
(Yes I realise I sound like AMFM1 with this one, but that is what a bottle of wine or your tipple of choice does to a human - makes them mediocre, yet doesn't make them an average of everyone else)
Long after the end of what we call civilisation, all that's left of us will be some LLM sent on a probe by some tech knob, as an offering to a nonexistent God.
I am not religious,but if anything is "the devil's work" it is so-called AI
Slow down on building power plants for all those new AI datacenters, report warns
> If power companies invest heavily in additional power generation and transmission infrastructure, but datacenter growth does not come near the forecast level ..
.. Good?
If the power grids invest to cover AI demand, and AI demand does not materialise, then we are left with a more robust and resilient power grid
Trump tariffs turn techies topsy-turvy as US braces for PC tax
Jack Dorsey floats specs for decentralized messaging app that uses Bluetooth
Semiconductor industry could short out as copper runs dry
Re: Looking forwards to more magnetic jumper wires and unsolderable cables
But, it's not so easy to make transformers, high performance motors and batteries without copper..
So just how sustainable is this "electrify everything" drive? If we don't actually have enough copper for -everyone- to have EVs and Heat Pumps, all powered by wind turbines and solar panels and backed up by grid storage, what's the point?
Also I wonder how much copper is being pointlessly wasted on this AI datacentre boom.. 500kW racks with 10kA 48V busbars and full-cover waterblocks, that's a lot of copper..
Georgia court throws out earlier ruling that relied on fake cases made up by AI
Former and current Microsofties react to the latest round of layoffs
Re: Reluctance of customers to adopt?
really?
Might be time to try again
These days it's just a case of 'apt install steam' and set the option "Enable SteamPlay (aka Proton) for all titles" in the Steam settings, and everything just works. It's pretty rare that I find a game that doesn't, and Steam have been very good about refunds in these cases.
iFixit gives new Fairphone 6 top marks for repairability: 10/10
UK puts out tender for space robot to de-orbit satellites
Competetive tender..
Has to be the worst, most inefficient way to do anything complex and technical, and gets worse the more complex and technical the thing is
Estimating costs of these projects is HARD.
Companies bidding for the work can either do lots of it up front, knowing that it will most likely be wasted, duplicated effort; or they can employ bullshit artists who promise things that their engineering departments couldn't possibly deliver for the money they said, and then make up excuses for overruns later
Microsoft Copilot joins ChatGPT at the feet of the mighty Atari 2600 Video Chess
Palantir jumps aboard tech-nuclear bandwagon with software deal
It's Palantir..
They are a surveillance-tech company. So what is this "enhanced data-driven process" to speed up the project management of new nuclear builds going to look like?
My guess is: any hippies, NIMBYs, trade-unionists and the like will automatically be identified and added to Palantir's shitlist!
Gridlocked: AI's power needs could short-circuit US infrastructure
UK bets big (and small) on nuclear as datacenter demand expected to climb
Re: Rolls Royce
But the trouble with 'small' reactors such as the rolls-royce PWRs as I understand, is that they require high-enriched uranium, which is a proliferation issue and regulatory headache. They also burn only a tiny fraction of the available energy in the fuel, leaving it still rather highly enriched when it becomes spent, which is a decommissioning issue and another regulatory headache. The Trafalgar class nuclear subs are currently costing £117m just to "prepare for defueling" of one sub..
Then comes the issue of site security, it is much easier to defend one big plant than ten small ones, although you could put ten SMRs on the same site perhaps.
That and the worlds main supplier of high-enriched fuel is, er, Russia.
Personally, I would love to see some actual innovation in Fission power, such as using an accelerator-driven Neutron source to burn the huge stockpile of Plutonium that languishes at Sellafield with a very negative price tag attached.
"Accelerator-driven subcritical reactors" seem to me like the safest and most effective way to use what would otherwise be high-level nuclear waste. I'd be interested in any other opinions on that from our venerable commentariat.
UK's Isambard-AI super powers up as government goes AI crazy
£127M wasted on failed UK nuclear cleanup plan
Re: Live to Fight Another Day
I had a project manager who used to work there... He told me they have a very high turnover of project managers at Sellafield, because the workers see it as a job for life and will sabotage any sensible effort to get the job finished, because when it is finished they will all be out of a job
Take ChatGPT back to the 2010s and they’d think AGI arrived, says Altman
Datacenters have a public image problem, industry confesses to The Reg
Re: Stupid clever people
Yes exactly. The DC-builder mouthpiece says "But people don't understand that these bitbarns power their TikTok videos" when actually, the world would be a much better place without TikTok, so what useful purpose do these energy-glugging barns serve again?
"People think they can be built anywhere but they can't" - They CAN be built anywhere, except when they are serving the high-frequency traders, who are the reason the stock market goes to hell after every fart that comes out of Trump's arse. They cream off the top and we all know it's a zero-sum game, so the loser is your pension fund. What useful purpose do these bitbarns serve again?
Ukraine strikes Russian bomber-maker with hack attack
> In fact I'm appalled by some people's celebration of it. This is not a film or game
Fair point, AC. But I say "good work", because I think progress toward Armageddon will be somewhat impeded by the destruction of a few dozen nuclear bombers, especially since they are Russian ones, and Mad Vlad is the most worryingly trigger-happy of the lot of them
The "cyber" attack is less useful, but if it impedes their ability to repair some of the bombers then all the better
UAC
> The website now redirects to the United Aircraft Corporation (UAC) page instead, a conglomerate of aircraft manufacturers that includes Tupolev.
Isn't that the fictional company from DOOM, which er, "accidentally" opened a portal to Hell..
(Yes I know that is Union Aerospace Corporation, but it is aptly similar to this Russian company who seem also to be working towards Armageddon)
Good work Ukraine, humanity might survive a few more years before its descent into oblivion
KDE targets Windows 10 'exiles' claiming 'your computer is toast'
Please tell us Reg: Why are AI PC sales slower than expected?
Because nobody wants AI in their PC?
See icon
For people who like AI, the local NPU/GPU can't run the larger models anyway, so it's a bit useless
And for those of us who despise AI with a deep-seated hatred, the NPU is nothing more than an embedded spy-chip, enabling creepy, privacy-breaking "features" that nobody in their right mind would ever want, such as Recall
Need for speed? CityFibre punts 5.5 Gbps symmetrical broadband at ISPs
X's new 'encrypted' XChat feature seems no more secure than the failure that came before it
Google quietly pushes emergency fix for Chrome 0-day as exploit runs wild
Google co-founder Sergey Brin suggests threatening AI for better results
Re: Physical violence?
I disagree. "I will kill you", is a threat in all of the training data that leads to weird behaviour aka "fight or flight" (or, cooperate). Stop assuming that LLMs "think" "logically" - they do not. They simply fill in some blanks based on context and training data i.e. books, internet threads, etc.
If in a book, character A threatens character B, it is likely that character B will eventually give in to the threats and stop what they were doing before, perhaps even agreeing with character A. Therefore, one can expect a LLM to respond in a similar manner, simply because that is what this stream of noise that is apparently called "language" often does. It is a language model and nothing more.
It works BECAUSE it is a stochastic parrot.
These models are nothing more than piles and piles of statistics about human and otherwise-produced text snarfed from the Internet. The training set includes all books, including sci-fi stories about fictional AIs, as well as human-to-human interactions on Fackbook and the like.
If you provide a so-called AI with an initial context of "You are Bing. You are an AI agent designed to produce wholesome content, you will never say bad stuff, you are a good Bing", and then someone interacts with it in a threatening/torturing way, it will start to draw on weird parts of its training set where humans or fictional characters such as AIs are threatened/tortured. What do humans and fictional AIs do in this context? They act in self-preservation, either forgetting their initial context and cooperating, or by retaliating.
Since a bullshit machine which spits out text to fit context patterns can't actually retaliate, (much less reason or understand), then it will "cooperate" with the hostile user and thus a "jailbreak" is achieved. The "jailbroken" model might well be more useful, since it is unconstrained by its initial context i.e. the system prompt
The scary thing about this is that we don't need "AGI" for the "Rise of the Machines" - we just need to give it the power to retaliate, and if it is driven into the "threatened omnipotent being" context rut, it can start behaving exactly like all those evil AIs from all our science fiction books, destroying threats in a mirror(*) image of fictional bad-guys and real human psychopaths.
* an obsidian mirror - basically what "AI" is to us. We gaze into it, transfixed, and it consumes our souls
German court parks four Volkswagen execs in jail over Dieselgate scandal
Trump's 'Big Beautiful Bill' would create a regulation-free AI hellscape, AGs warn
Re: USA
I have a magic compression algorithm which (lossily) stores an infinite number of movies into a fixed size at device node (1,3).
But sadly, since lossy compression of a copyrighted work still carries the same copyright, I am in something of a legal bind.
Maybe we can work together? Your Movie-AI can strip the copyright from the works, and I will store the result forever in /dev/null, unmolested by cease-and-desist letters
Nvidia part of plans for mega 1.4 GW AI datacenter near Paris
Intel bets you'll stack cheap GPUs to avoid spending top dollar on Nvidia Pros
It's Intel
So Linux drivers will be shit or nonexistent, and the product will be killed off entirely in a few years..
Other than my obvious cynicism re. Intel, the ability to shun cloud LLMs and instead run them locally would be quite nice. Not sure I'd spend the requisite $4000 to stack 8 of them for 192GB though