Posts by MonkeyJuice 249 publicly visible posts • joined 4 Nov 2016
In Apple's favour here, say someone is opted in, and a friend emails them who has not opted in. Just the act of sending an email to a trusted address has now violated their privacy expectations. AI is a giant sponge that soaks up everything around it indiscriminately.
She posted on X that Gen Haugh and his deputy Wendy Noble, who US media reported was also let go, "have been disloyal to President Trump. That is why they have been fired."
Before their firings were reported, Trump told reporters he would get rid of any staff deemed to be disloyal.
"We're always going to let go of people – people we don't like or people that take advantage of, or people that may have loyalties to someone else," he told reporters aboard Air Force One.
That's not speculation, that is a statement. Maybe read what idiotic shit they've been taking credit for before you attempt damage control.
I have a friend who's a bit of a novice and wondered why his shiny development M1 laptop was thrashing hard. On closer inspection it turned out this fairly recent machine had 8gb ram in it. Explaining that I had double that on my GPU alone, on a rig that cost half the price, a lesson was learned about the cost of poor hardware choices.
This statement also isn't really novel, just a weaker form of abliteration.
I'm starting to suspect that ML peeps don't hold themselves to the same standards as the CS folk.
Mine's the one with the five sigma.
Or my favourite recent bugbear, Geoffrey Hinton, stating multiple times on news broadcasts without pushback, that he believes AI is already 'conscious', despite there being absolutely no fucking objective way to tell, and absolutely no non-circular argument as to why it should be. Since he's today's "Godfather" of AI, he's now also apparently a tenured theologian and philosopher that believes in sentient toasters, and we must all get on board.
Wayland's nuking of XInput pretty much nerfs reasonable tablet support.
And as gtk-4 slowly rips out basic things like "place a window at this coordinate on this display" to bend to Wayland, the UI experience is slowly but surely becoming worse and worse. "Hey you can add that back with a gnome extension" is not an answer.
Organisms don't evolve. Only populations can evolve. The fate for individual organisms is death.
This is patently false, and has been well accepted for the last 30 years.
This isn't a new discovery, so there really isn't any excuse for blathering this tripe on here.
Please don't make blanket statements on the internet when you don't know the subject, it just lowers the internet's IQ even further than it is.
Correct. Two people in this thread, including me have had positive experience, and I have multiple friends who have similar. I'm merely sharing my experience, not broadly describing the state of affairs universally. You could have just said "Not in my experience" and not come across quite as contrite, but you do you I guess.
I either have to VAT register, or take a few months off each year, depending on how I feel, so go figure.
Your singular poor experience doesn't reflect the reality of others, and none of the projects I contribute to are nodejs or php projects, if that helps...
Adding better test coverage to a particular feature set would be something a lot of projects would benefit from and appreciate, without requiring a huge investment, for example.
Large, well managed projects also often have a "good first issue" tag, which can help people gain familiarity as well.
I think the problem to a certain extent is, particularly with new developers, there is no real guidance to how any of this works, and it will often require reaching out to the developers directly to find out where to begin, which instantly discounts a lot of people nervous to do so.
Absolutely disagree. I contribute to several projects and this regularly sends work my way. If you actually know a real world system inside and out, there are people who want to leverage your knowledge to help introduce it into their work.
Contracts like these tend to work as part development / design, part training developers within the company on the software / mindset of said project.
Or it could just be you got lucky.
You'll need to repeat this experiment at least 1000 times with careful review, at which point you could draw a vague conclusion. Three sample points are not adequate to draw anything from. If you find it performing perfectly each time even then, I would eat my hat.
"Whatever's going on behind the curtain, it's doing more than echoing it's training data."
Sorta sort of. It's a huge tangled ball of wtf. Things like word2vec showed you can do some interesting algebraic 'reasoning' from unsupervised training at the word level. For example (man - king + woman == queen), which actually makes sense when you look at how it works.
At the scale of these tens of orders of magnitudes larger models, nobody really knows why it is doing anything, and some portions of the network may be simply undertrained but we don't know because we've not tried it yet (and will never have enough data to satisfy training it anyway, or adding more data will cause previously working parts to start barfing). When it comes to anything that requires auditing and accountability, this should be an enormous red flag.
The problem is we've had things like this relying on far, far older technology since the late 90s, too. Sure it's slightly more accurate now, but a move from 97% accuracy to 99% accuracy is really only an extremely minor improvement in the grand scheme of things when compared to an actual human.
When it works, it feels magic, and it may work a lot of the time of the time, but to prevent something incredibly stupid from creeping in you have to just sit there, validating every output. Fatigue sets in, and even the human reviewer lets things slip through the net.
While we have made impressive advances to hardware, and transformers are a considerable improvement on architectures that have come before, we are simply not there yet, and nobody really knows how to fix these issues. Despite what the AI bros at large companies may say, academia is considerably more cautious unless they're pitching for funding.
"So defence mis-spends and overspends increase GDP, but produce nothing"
Not quite true. They can occasionally produce long queues at the hearing clinic, for example.
I see you have been terribly let down, which is I suppose why screaming on the internet has become your main outlet.
I guess all that is left to say about your rage fuelled antagonistic posts is "You won. Get over it"
Maybe think before popping that X in the box next time, or you'll end up with the world you deserve rather than want, eh?
It's been both entertaining and enlightening.
Keep on winning, and toodle pip!
"Farage did a good job getting the brexit vote"
I feel we are approaching the nub of the matter. What, prey tell, are the Brexit benefits you are happiest with? I seem to remember Moggie had to form an entire department to canvas the entire UK population for one.
Our newly invigorated trading terms with the US, perhaps?
The amazing and immediate cessation of illegal immigration?
The fact the EU now has to bow to our every whim, because we are no long invited to the meetings?
Ephemeral 'taking back control' nonsense need not apply here, since I am asking for something nice and tangible.
What did you want out of it, and what did you get?
Curious. I'm guessing you're more of a Farage guy? Or PA perhaps?
It seems odd to be crowing so loudly and so consistently. Maybe you're Grant Schapps having a little time on the interweb?
Still. It makes your poor handle on the US political system far more understandable, I have no other idea why you would be looking at that shitshow and thinking 'gee, this is a great idea, more of that please'.
Speaking of relevance, do you even code?
Trump doesn't affect me because he is far, far away, in a country that is increasingly becoming irrelevant. It's a good old laugh to tune in to some of the wailings from the far right such as yourself from time to time mind you.
I do love a good English breakfast though. I have been recently having an extra fried egg, as a little treat. One for my homies, as I believe you say on the other side of the pond.
Keep on winning!
All the best.
The fun part about software is that it's completely unregulated. We've been screwing things up at scale for 70 years now, and we're still skipping regulating the software side of things and moving straight on to vague 'AI regulation' instead.
The advantage of Civil Engineering is when you guys catastrophically screw up, everyone knows what you did while they're digging through the rubble, and even the great unwashed understand the need for oversight.
What I find hilarious about this misappropriation of the 80/20 rule to a language model of all things, is that you have no idea where or what that 20% is without a fine tooth comb, and when it is wrong, it isn't "good enough", but catastrophically wrong.
The C-Suite regard this as "I can fire 80% and the 20% will pick up the slack!".
Not that I would want to stand in the way of the inevitable march of 'progress' here.
While my answer was hyperbolic because I am not the joyless soul you are, the logic is pretty straightforward and my underlying point remains, so I'll explain it simpler for you to save those tired fingers of yours:
An AR15 on a modern battlefield is absolutely useless, which is why the US Army does not issue AR15s.
It is, however extremely effective in a shopping mall or school.
The dems don't want AR15s because of the latter, and the reps let you keep your giggle guns because of the former.
See? Simples.
Now give me my crayons back I'm hungry.
17. President Trump had authority and responsibility to direct deployment of the National Guard in the District of Columbia, but never gave any order to deploy the National Guard on January 6th or on any other day. Nor did he instruct any Federal law enforcement agency to assist. Because the authority to deploy the National Guard had been delegated to the Department of Defense, the Secretary of Defense could, and ultimately did deploy the Guard. Although evidence identifies a likely miscommunication between members of the civilian leadership in the Department of Defense impacting the timing of deployment, the Committee has found no evidence that the Department of Defense intentionally delayed deployment of the National Guard. The Select Committee recognizes that some at the Department had genuine concerns, counseling caution, that President Trump might give an illegal order to use the military in support of his efforts to overturn the election.
You can find out why that happened in the findings of the Jan 6th committee. Quite easy to turn a bunch of used car salesmen and rednecks into an issue when you refuse to authorize the national guard, innit.
Why do you suppose Trump would do that when it isn't in his favour?
OPM data includes the security clearance of employees. This data is NOT classified confidential, particularly when it's everyone on Uncle Sam's payroll.
This data contains information which could expose very Republican leaning personnel, endangering life and limb from foreign and domestic threats.
The problem, I think, is you are still under the illusion that the Republicans are the one's running the show.
Something about closing the barn door.
The data is already exfiltrated to Azure. Who is auditing the proper disposal of data? And what about the fun and exciting code additions to the OPA systems, are they all carefully accounted for and able to be reverted? Are they still in the building? What is the status of the backups? Where is the mystery DOGE server currently? It is one thing writing an angry letter and another actually mopping up APTs.
Time will tell, but I am absolutely not holding my breath.
What a security clusterfuck.
People are working on a proposal for C++. However, there are a couple of pertinent issues here:
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
