[ b / int / meta ] [ fefe ] [ test ] [ FAQ / Rules / HowTo / Stats ] [ Radio / BNR ]

/int/ - certified time wasters

Subject
Comment
File
Password (For file deletion.)

 No.15 [Last 50 Posts]

I bought a new (used) ThinkPad which came with Windows 11 pre-installed.
I never had a problem with Windows but i played with the idea to go full Linux on this ThinkPad.

What are some good distros these days for people that don't want to fix stuff all the time?

Also: Computers general

 No.18

ubuntu.jpg (27.89 KB, 1044x785)

linux mint.png (2.97 KB, 225x225)

>for people that don't want to fix stuff all the time?
I thought that was the whole point of using Linux. Joges aside: As far as I know it's still Ubuntu and Mint which are easiest to work with. Not a big fan on Linux for private computers though, I tried it a few times, and sure, you can make things work, but if I calculate the necessary time a noob like me needs need for fixing issues in comparison to simply buy a Windows licence, Windows suddenly looks pretty cheap.

 No.28

Ubuntu, thus also Mint, just started with turning into non-free, pay per use, distro.
Just stick with Arch or Debian.

 No.29

If you want bleeding edge Arch if you need solid stability Debian. Both are very simple to install and take about the same time to fully set up after the initial install like win10/11 to get everything to your liking.

Arch
You simply enter "archinstall" into the terminal after boot and follow the instructions
Debian:
You likely want to get the proprietary drivers edition for compatibility reasons else just follow the GUI

 No.99

Kubuntu, Xubuntu

 No.107

Manjaro works fine for me.

 No.108

linux mint is good it's just werks distro
btw every distro working well unless you install meme distros like guix gentoo void etc.

 No.114

>>15
>distros these days for people that don't want to fix stuff all the time?
I like openSUSE Leap for that reason. Unfortunately, it's about to be kill. Maybe Fedora or Debian Stable.

 No.133

try popOS, it has a business behind it so its maintained

https://pop.system76.com/

 No.135

>>99
>>108
See >>28. This is really not the appropriate time to recommend Ubongo based distributions. People who are just trying to switch don't want to switch in 6 months against because Ubongo has literally become Microsoft Linux with online accounts and payment.
Yes I'm still using Kubuntu on my main machine because I'm lazy but it's not much work to switch to Arch/Manjaro or Debian for that matter as soon as they cut their free cable. For someone who is planning to "get into Loonix" it might be more of a hassle.

Not sure if I read that correctly from OP. >>15 Are you experienced with Linux?

 No.296

hi.png (304.78 KB, 601x443)

>>15
Update: I installed Mint.
I got my first error message when trying to install. A couple more afterwards and then i already gave up. Nothing just werks.

Still, i wasn't interested in Windows 11 so i installed Windows 10. Comfy Windows 10.

 No.299

>>296
Looks like you can be nostalgic about anything as long as there is a successor you hate.

 No.389

word error rate.png (77.24 KB, 886x1164)

Why do you think transcribing speech (with OpenAI's Whisper) produces such different error rates between similar languages? Smaller is better.

Think I'm going to try it out myself soon. Too bad I don't have enough vram for the large model though. Also if anyone happens to know anything about speech-to-text, specifically doing it locally, please share your wisdom.

 No.391

>>389
I guess, because there is much less data for Belarusian then for Russian. Even if it's still a lot, when a model hears East-Slavic speech, it's statistically safer to presume that it's Russian and process it accordingly (with different grammar rules).

 No.392


 No.393

Not locally:
Kaggle provides 30 gb of ram in cpu-only environment and 15.9GB gb of gpu memory.

 No.404 KONTRA

>>391
Right. I wonder given equal training data which language would fare the best. Which would be the language of the robots given that they had to use a human language to communicate verbally?

>>392
Have you used any of these? Particularly interested in multilingual ones. T. Hanks.

 No.405

>>404
> Which would be the language of the robots given that they had to use a human language to communicate verbally?
Arrrkchually, phonetic is just one aspect of simplicity. English is very primitive language overall, but its phonetic is complex.
https://www.youtube.com/watch?v=A8zWWp0akUU
On the other hand, there must be a language with 100500 tenses and genders and rules, but which still "is written same way as it spoken". Maybe Finnish?

> Have you used any of these?

No.

 No.408 KONTRA

And within phonetics, speech recognition is only one aspect. For example, in Arabic there are only three vowel letters: one for a (like in man or like in hard), one for "u" and "o", and one for "e" and "i". It's easy to write a speech, you won't confuse e with i, but on the other hand when you try to read such text, you don't know how exactly to do it.

 No.418

>>15
With an old ThinkPad you are already off to a good start, but you still need programming socks and cat ears. And dont forget putting random stickers all over it, as well as telling everyone near you, that you are using linux.

 No.541

qqqqqqqqqqqqfdsfdssfddsf.jpg (164.11 KB, 1080x1081)

qqqqqqqqqqqdsfgsfddfggsfd.jpg (69.64 KB, 750x400)

qqqqqqsssadffdfsafds.jpg (88.25 KB, 1082x544)

ML text generators are just more granular search engines.

Any "understanding" that you see in the generated text is just a projection of your own understanding into structured noise. A form of textual apophenia.
The illusion breaks down when you tap into a very high confidence probability space, but change 1 crucial detail. The text generator will respond as if you asked the high confidence thing, and won't "notice" the small, crucial detail.

Unfortunately (fortunately?), metaphysics and epistemology are not yet "solved".
"Knowledge" or "understanding" exists in an ambiguous duality between the information contained in thing itself and the information that the observer uses to "decode" the information.
You need context to parse information, and you need information to exist within a context in the first place, but it's hard to tell what, and in what proportion, contributes to the "meaning": the context, or the information itself.

So to attribute "understanding"/"knowledge" to any singular entity, rather than viewing it as a relation between things, is a form of naive neoplatonism.

 No.542 KONTRA

>>540
> The text generator will respond as if you asked the high confidence thing, and won't "notice" the small, crucial detail.
As well as human would.

 No.543

>>542
Speak for yourself, I'm all about details.

But one thing I agree on is that there are different "modes" of mental "presence", and the state of being disinterested and distracted is, indeed, very "mechanical", and does indeed occur in real humans
Like listening to someone talk about something you don't care about, or a boring lecture, is similar to how an NN would work: you "notice" the familiar / high confidence things, ignore the in-between details, and re-construct meaning from those disjointed pieces you remember.

The thing is that, all of the institutions of "understanding" throughout human history implicitly recognize this, and are built to maximize and utilize the mental "mode" of being at maximum attention and carefully considering every detail, rather than extracting the lossy average out of the things presented.


The problem with NNs is then, that they replicate human behavior at its dumbest, rather than at its best.
It is a mechanical dumbfuck simulator. We have enough dumbfucks, already, what we really need is something that simulates the low probability, but "correct" person.

 No.544 KONTRA

I dunno mang. When you speak or perhaps even write, you have a general idea what concepts you're discussing but the actual words that you end up using come up from the noise. You can't tell me that you source and rationalize each and every word specifically. Coming up with rationalizations for the words after the fact sure, but that tells nothing about the reasons you came up with that word in the first place.

Whether or not current AIs even have any sense of the concept is another thing. Not sure about that.

But you reading this text is essentially a projection of your own understanding projected onto my noise-vomit. And no, this wasn't an AI generated gotcha-post even though it might seem like it.

 No.546

Perhaps the next stage for "AI" would be some kind of generalized concept of "introspection".

You see, it's true that humans memorize general "concepts" and then relate them to each other loosely through some likely "probabilistic" scheme, but humans don't JUST do that. Humans also take the information they already have and "process" it into some new relations, or generate second-order, "new" knowledge from information already gathered. (Isn't that the current theory of what dreams are for?)

So, for AI to not just be an easily tricked probability machine, the next stage would be to take all the information it already gathered, and using that, INFER some further general "relations" or "systems" that don't obviously map to language or "obvious" things, but form a sort of network of abstract relations.

It would have to get more and more abstract with each stage, until pure "data" can be contextualized inside some kind of "system" or "model of reality".

Maybe it does all come down to a number of connections per "node". Like, the reason I know that 1 pound of feathers is not equal to 2 pounds of steel is because there's some connective pathway from that specific question, to my notion of what "weight" is, while the neural network doesn't have enough connections to form that pathway. (Or maybe it lacks the capacity to form abstract notions in the first place?)
But I imagine that the way to form that pathway, is not to merely gather more data, but to make new connections between the data the NN has already gathered.

 No.547 KONTRA

And what about communication across senses?

Using as an example the myth of Newton getting the idea for the theory of gravity from seeing an apple fall from a tree, isn't an AI severely deprived by its inability to experience "sight", and "motion", and only being able to think in terms of "words" and relations between words?

And what about sound?
And what about the invisible connective tissue between all the senses, where an apple makes a "thud" sound when it falls on grass, but a different sound when it falls on stone? There's information there in this "between" space, and how could AI ever learn that, if it's only seen apples, and only heard "thud" sounds, but never heard a "thud" sound from an apple falling on grass?

 No.548 KONTRA

Yeah I was thinking about the sensing part too when it comes to the AI-weight-retardation. Things we can sense ourselves are a lot easier to conceptualize. As soon as you step away from that, even people start struggling. Compare the understanding between something like infrared and the visible spectrum. Understanding the former is probably something not everyone can even achieve.

>the way to form that pathway, is not to merely gather more data, but to make new connections between the data the NN has already gathered.

Probably.

 No.553

>>543
>The problem with NNs is then, that they replicate human behavior at its dumbest, rather than at its best.
The biggest problem with NNs is that some people think that it's very cool to be a resentful grumbler. Few years ago idea of electronic bydlo seemed impossible. And now you are unsatisfied because it's not an electronic Newton.

NNs are able to memorize very specific patterns and to reply according to them (same as people, especially bad students at exam). But this doesn't mean that they're not able to generalize and to extract abstractions behind data. An example of how they do it on a tiny network and a simple task:
https://cprimozic.net/blog/reverse-engineering-a-small-neural-network/


Btw, Newton was wrong same way as NN.

NN thought "in questions like that answer is equal because it's always like that in data". Then brick provided it with a different data.
Newton thought "acceleration is always proportional to force because it's always like that in nature". Then observations of Mercury's movement provided a different data.

Both is inductive knowledge, and difference is merely quantitative. In this sense there is no "real"(tm) understanding.

 No.555

>>548
I've worked with hyperspectral cameras for four years in my first job (paying for uni) and I can attest that nobody in the team was able to build a useful intuition of what a hyperspectal image is.
We could build intuitions about single bands easily, for example the frequency of our NIR cam that was within 2nm of the spectral bands of chlorophyll (and could be used to detect vegetation) was quite easy to intuit about even though our eyes only "see" this frequency as a product of three other frequencies.

But an intuition for the entirety of spectra within one image was not obtained - where intuition means the ability to intuit (magically reason) about it in such a way that you don't have to think slowly and with concentration in order to understand something about the data.
We would visualize the bands of a single image in many different ways, but none made the whole of one image "make sense".

It was an interesting and somewhat humbling experience in retrospect. I'd love to upload one of those images here but they were all classified and I had to delete them when the job was done… kinda regretting now that I didn't keep at least one data set and my own code.

 No.585

virilio_vision machine1.png (346.7 KB, 1780x1262)

virilio_vision machine2.png (814.41 KB, 1772x1255)

virilio_vision machine3.png (247.9 KB, 955x780)

>>548
>>555
I don't understand that second post in its technicality but I had a seminar on technological sensation including remote sensing the seminar was not philosophical nor technical (obviously not the latter) but quickly branched out into the politics of technological sensing, like drones and satellites in war and open source intelligence, various forensics and I had struggled with machinic sensation as concept and reality because humans are just too used to interpret it in their physiological way. Machines or "satellites and their cameras" don't see, they sense and they do sense in a way that is hardly comprehensible. Quite fascinating. There is an old text from late 80s by Virilio which has some nice sentences in it that bring it down. the rest of the essay is rather uninteresting, didn't finish

 No.688

>>585
That was super interesting to read.
Virilio seems to be attempting to evaluate the technical aspects without a clear understanding of digital computing. His argument about us humans not able to see what the machine is interpreting only makes sense at a time before images could be manipulated and copied digitally.
I'm referring to the line

>Having no graphic or videographic outputs, …


in the second image.

I might be misinterpreting something though, because I lack the context of his writing.

>I don't understand that second post in its technicality


Ask away if there is something specific.

 No.1221

Am I dumb or did Chrome have a function to download stuff with alt-click without opening an extra window?
Now when I press Alt and click the download window opens instead of downloading that shit directly.
It's annoying.

 No.1223

>>688
>His argument about us humans not able to see what the machine is interpreting only makes sense at a time before images could be manipulated and copied digitally.

So do machines sense like humans do physiologically or is the digital sensing of machines equal in its quality? Because that is how I understand it, maybe I'm extending Virilio here but as far as I know, what the machine puts out is converted to something that humans can perceive as image, so what we see as output is not what the machine actually sees but a conversion for humans. Maybe I'm wrong though.

 No.1224

>>1223
and I guess the machine "seeing" something is troubling. What we see as output and what the machine senses of its environment are not the same. That is the question and what I think is what Virlio can be extended to. Because it makes sense that in 1988 he did not think of the machine giving an output. I overread that because for me the other question is more important.

 No.1228

>>1223
You need to ask yourself: what distinguishes "seeing" from "non-seeing"?

From functional perspective, NNs are definitely able to distinguish objects on images and to act on this behalf.

How it works?
RGB-encoding for images in computer is inspired by cone cells in retina. Convolutional networks (alpha and omega of computer vision) are inspired by neural networks of visual cortex.

"Inspired" doesn't mean "work exactly the same way". But some insects with facet eyes process visual information perhaps even more differently from us, and no one doubts that they are able to see.

So yes, I'm sure that they can see.

 No.1264

im this cat 29.jpg (24.36 KB, 568x377)

What's the purpose of building (multiple) satellite networks for high speed internet connectivity?

SpaceSex has Starlink, Amazon is building Kuiper with similar characteristics, and there are some others too. Seems an awfully expensive effort to provide good connection to remote areas, marine vessels, etc.

The hardware required to receive and transmit data to satellites is going to be a lot more expensive and energy intensive on a large scale, than it would be to build a traditional network in third world countries without proper internet, or so I'd think. Same goes for direct satellite internet connection for phones, it's going to use a lot more power and battery capacity isn't there to support it. Is the answer Burger national security?

 No.1269

>>1264
It's faster and safer, especially in war situations. wire is critical infrastructure, you can have for example redundancy through satellites
If you do not have private enterprises but states and their satellites program in mind, that is for security and autonomy reasons. Many states have no satellites services and need private "help", so profitable business for Musks and Bezos of this world.

For Germans interested in the role of satellites these days:
https://www.deutschlandfunk.de/internet-von-starlink-und-co-satelliten-technologie-in-krieg-und-frieden-dlf-caff66db-100.html

 No.1277

>>1264
Hint hint: The US military financed Starlink to a large extent.

 No.1293

Let's say I want to keep a very simple database, just a tick above excel tier, what is the best free program I can use?
OpenOffice Base wants me to install JRE, how is Libreoffice? MSOffice is out of the picture here, as is anything with MySQLServer/SMSS etc.

 No.1295

>>1293
Sqlite + some gui for it if you need it.

 No.1296

>>1293
Google sheets with public access + python interface

 No.1426

three letter agent.jpg (40.21 KB, 600x608)

Apparently Snip & Sketch, W11 snipping tool, cropping tools and some phones have a bug that when you crop and save a picture, it saves an entire screenshot or the original image within the file.

I use snipping tool daily, in many cases to redact important information, but as far as I can tell W10 snipping tool isn't effected.

More information somewhere around here, I don't fucking know how Twitter works:
https://twitter.com/wdormann/status/1638235267919233024

 No.2201

1.jpg (117.06 KB, 618x770)

2.jpeg (120.18 KB, 1080x1086)

Let's see if I can navigate through aws 12-month "free trial" without getting fleeced by Jeff Bezos.

 No.2212 KONTRA

admin kat.jpg (48.57 KB, 496x370)

>>2201
Please do report back on how it works out. Can you share what you'll be using it for? And please don't name your instances/services as "orja"s. That's against code of conduct, mmkay :DDD

Good luck! Interesting stuff.

 No.2215

Is it possible for a human being to write correct code on modern C++? Probably for grandpas from the committee with 50 years of experience. Few more standards and they'll pass away it will require an artificial intelligence.

 No.2244 KONTRA

>>2212
Proxy server. I mostly want to make use of their CDN.
Depending on the traffic usage I'll consider hosting some other services next month.

No, I won't give offensive names. My go-to shit variables and throw away names are test, tst, temp and tmp. Haven't caused any injection so far.

 No.2296


 No.3331

FIFINE A6T.jpeg (110.75 KB, 538x550)

FIFINE T669.jpeg (42.54 KB, 473x550)

Maono AU-A04.jpeg (34.01 KB, 497x550)

So, I've been thinking of buying a mic for gayming, and the obvious choice would be to buy a headset, but the problem is, I kinda hate headsets, so I looked at some cheap-ass "streamer"-type mics. Do any of you guys have experience using similar mics? I'm mostly interested in the possibility of using it together with speakers (not too powerful or loud, just cheap 2.1 system). Supposedly these mics are cardioid, so they should only pick up sounds right in front of them, but I'm not really sure. Will there be any feedback noise, if the mic is about half a meter away from speakers? Do I still have to use headphones with such a mic?

 No.3332

>>3331
My gf has a mic like this and it picks up my speech and keystrokes from four meters away. I’d guess sounds from your speakers might bounce around and get picked up too. Maybe there’s software to deal with that.

 No.3344 KONTRA

cardioid.png (188.77 KB, 700x693)

>>3332
Were you sitting inside its working zone when it was picking up the noises? And yeah, sounds bouncing off of the wall behind me could be a problem (my room is pretty small), so I hoped that these mics allow some sort of fiddling with their sensitivity (or signal-to-noise ratio, or whatever), so that they pick only sounds about a meter away from them, filtering out anything else. That is, I hope that they allow such fiddling, but I'm not sure, that's why I'm asking if anyone had hands-on experience with them.

 No.3353

>>3344
Yeah

 No.3367

ftcjnyto1alz.jpg (210.54 KB, 795x1032)

>Thinkpad
Cringe.

 No.3369

Imagine how much more useful ai chat bots could be if they didn't try to make them sound like human conversation for PR/ideological reasons.

 No.3370

>>3369
What else could AI chat bots sound like if not human conversation?

Also, it's not PR reasons, but technical reasons. Chat bot learns on human conversations - sees examples of natural language and extrapolates them.

 No.3371

>>3370
The actual "killer feature" of such chatbots is that they're REPL command line environments for arbitrary corpus of data. Or you could think of them as search algorithms, but they don't search for directly stored data, but for arbitrary patterns.

To shareholders, they're presented as "google killers" because they (purportedly) do the same thing as google search does, except they give the results immediately, without the additional step of the user having to click on a website link. For them, the benefit is that they don't have to pay ad money to websites any more, and it's as if google saved the entire internet locally and started serving you that. All ad money goes to them.
On the user side, the benefit is that you cut out that middle step of picking the search results, reading and parsing the information, etc.

I'm interested in the latter. The user convenience side.
So, there is no reason why you wouldn't train AI on ANY kind of data, including synthetically generated data sets, to have an interactive command line that gives you quick results for data that didn't had to be manually structured.

It's a general truth that tools become more useful the more specialized they are.
My prediction is that AI will become actually useful when people will start turning them into "fuzzy search engines" for highly specialized workloads, rather than pursuing the bullshit goal of building a "knowledge machine".

god, why do I have to explain all these things that are completely obvious to anyone who follows this shit it's like none of you people do your homework.

 No.3372

while these absolute rubes talk about the "existential dangers" of rogue intelligences that run on silicon, they fail to see that we already are existentially threatened by a rogue intelligence, but one that runs not ONLY on silicon, but on brains and paper and laws and guns and blood and in short, lives in ACTUAL PHYSICAL REALITY and occupies the unaccounted, invisible meta-space of emergent human behaviors
this rogue intelligence is called CAPITALISM

 No.3377

image.jpg (11.46 KB, 567x104)

image (1).jpg (73.04 KB, 1280x531)

>>3371
I don't know what exactly do you mean by "REPL command line environments for arbitrary corpus of data", but doesn't all intellectual activity (besides initial goal-making) falls under this definition? The corpus of data is your life experience (including books and lectures) and the command is whatever your boss or your client tells you to do.

> It's a general truth that tools become more useful the more specialized they are.

With LLMs it's not true, because connections used for one task can be reused for another task. Different skills are synergistic. It's probably the same for humans. If you have degrees in math and chemistry, you'll learn physics better and quicker. so it's more like a human than a tool... really makes you think...

As for search engines, they also use machine learning. And you don't need an extra click, now they show you a processed result on a search page.

>>3372
Good punchline.

 No.3378

> god, why do I have to explain all these things that are completely obvious to anyone who follows this shit it's like none of you people do your homework.

I can imagine shit which you follow

> shareholders are being told...

> it saves internet locally
> a glorified pattern matching
> muh tech bros

When you read it pretentious round glasses materialize in your mind. Internet leftists are bitter cunts about everything ("critical theories"), but AI is an object of special hate because one of the few things it does good is writing essays, so it's taking bread from their mouths.

https://the-crypto-syllabus.com/yanis-varoufakis-on-techno-feudalism/ - look at this one, all the text can be generated from seed "write an angry leftist interview about monkey pictures, add occasional praising of CCP" (and probably was generated from such prompt, but by human), it doesn't contain any specific information besides that.

 No.3379 KONTRA

>>3377
>so it's more like a human than a tool... really makes you think...

While it is a sort of model of what happens neurobiological the debate if everything can be done this way to emulate human intelligence is not settled but hotly debated.
And even outside of it a bit of critical theory wouldn't hurt you as it is a pleasant change of perspective when it comes to understanding the debates around AI itself because usually the business/applied science crowd is doing circles for decades when it comes to insights. They churn on the same questions instead of asking themselves what they are doing by doing so and if they might ask the "wrong" questions.

 No.3380 KONTRA

>>3379
You know people in the AI field and all of these people being feared of hype-fascinated are a bit like different philosophical positions bringing forth the same arguments and problems again and again, there is no progress in sight. The basic ideas for what we can experience today are quite "old", the computing power is better, though. But given that you need a shitton of energy to do things a human can do with way less I have my doubts that we are anywhere near "cracking the code" itself a nice phrase to ponder about because it seems to presume what has to be shown to be the case first of all

 No.3381

>>3380
>But given that we need a shitton of energy
For practical purposes, that doesn't matter. Energy in sufficient amounts can be provided by exploiting common goods, and if people who own the world decide they need it, they will find a way so it will be provided to them at near zero cost. Maybe they will irradiate us all and turn everyone into mongoloid retards with leukemia or maybe they will immolate the planet, but if it is technically possible to substitute labor with capital, it will be done.

 No.3399

>>3379
I have nothing against humanities in general, but so far I see clueless and biased people, living in their own bubble (which I'm guilty of not following close enough) and making arrogant claims without knowing basics of the subject.
> the business/applied science crowd is doing circles for decades when it comes to insights
Maybe, but how can these people help here? "I don't know what derivative is and not even interested in it, but everything in the world sucks, except China which is based for banning tutoring" surely won't help to come up with breakthroughts like this
https://arxiv.org/abs/1706.03762
(which don't exist ofc, tech bros just run multi-layer perceptron on big computing cluster, it saves internet locally and that's it)

>>3381
Good post. I can imagine the other German first agreeing with you, then stumbling upon "mongoloid retards" and fussily thinking "damn, that's a schizo, no-no-no, I don't support this".

 No.3724

Apparently the latest internet news is multiple social media companies fucking over their userbase by trying to monetize their services in various ways. Twitter was first with their paid checkmarks and charging money for API calls. Reddit also locked down their APIs and made them paid, and now Twitch is basically banning any sponsorships / ads that don't go directly through their platform.

The prevailing opinion seems to be that all these social media companies have been coasting on zero interest rate investments and running on no profit in hopes of outspending their competition, establishing a monopoly, and then monetizing once they're a monopoly.
Well, now they're monetizing.

And everyone is angry, and the sentiment seems to be "heh, you thought you could cheat the system with your unsustainable business model, driving out the competition while making 0 profit, well fuck you, I'm deleting my account >:) serves you right".

https://invidious.tiekoetter.com/watch?v=AdE-T8h8Xt0
Watched this video and had to laugh about the fact that he outlines all of the above and his ultimate conclusion is that Twitch had to be "respectful" and "be honest with the userbase" or some shit. Like ok, and what if they were? How does this change anything about the situation? Being "respectful" isn't going to change the fact that social media can't be free and make a profit at the same time.

Well my question is, was it a BAD thing that anyone could upload a video on the internet for free, and have it stay up there for anyone to view forever? Was it a bad thing that anyone could broadcast live video any time they wanted for free, and have anyone freely watch it? Was it a bad thing that anyone could make a forum just by clicking a button for free, not having to rent a server and install MyBB?
For all the faults of social media, I personally think that it was a Good Thing (tm) that you could do all these things.
So now that the investors are demanding profit, and preventing people from doing all these Good things for free, what the fuck are you cheering about so smugly satisfied? That the company now has to "play by the rules like the rest of us"?

As far as I see, you just lost free access to a service that was previously free, the once unified services everyone used will shatter into a million competitors, reducing connectivity, and the only ones who will actually benefit are "le investors".
And all because the "profit incentive" button was switched on. Maybe the profit incentive is actually le bad????
Maybe subsidized infrastructure is good for society and introducing the need to generate profit makes everything worse???//
Maybe the point of funding services is to have these services, not so someone can make a profit on them makes total sense?//??/

The internet can't be monetized. Central planning rules markets drools. The only question is what kind of central planning you want, white man: digital feudalism or glorious communism? Kill t he bourgeois class war now

 No.3727 KONTRA

>>3724
That was the model of most tech meme companies. Netlix, Amazon Prime and Disney were banking on the same thing, only for the market to solidify, leaving most of them in a deep loss, and suddenly having to pay 3x20 bucks a month for movies is not really a good deal.

Netlix is now back to releasing some shows by the week, ending the famous "binging model" and we're back to cable TV except it's a whole lot gayer and blacker.

 No.3730

>>3727
Reactionary zoomer

Central europe is like the north lands in warhammer fantasy, except instead of chaos energy, it's nazism

 No.3732

>>3724
> social media can't be free and make a profit at the same time
They aren't free, but you aren't consumer, you are a product. Consumers are companies which buy advertisements there. 8$ per checkmark are probably pennies compared to total twitter's revenue, and are more like a "fuck you" gesture to j*urnos and other blue checks.

Yes, it makes sense for internet platforms to be treated like roads or parks (common infrastructure). But tbh for big and important companies profit is not prime stimulus. It's like "private space corporations" or "private military companies", CEOs of such things are functionally not different from government workers.

 No.3733

Also, how would you centrally plan internet? Twitter is such a disaster, its format is suitable only for hysterical schizo oinking. Worse than QUERTY keyboard, historical anomaly. But it still should exist, together with Tiktok as ghetto for retards, to keep them away from normal people.

Good social network format is something like ≈2014 vk.com . And for serious discussions in topics something like LiveJournal .

 No.3819

>What are some good distros these days for people that don't want to fix stuff all the time?

Devuan. https://www.devuan.org/

 No.4337


 No.4510

1200x675.jpg (193.99 KB, 1200x675)

This unicorn startup faked 95% of its users:
https://www.youtube.com/watch?v=M_wZpSEvOkc

This Prigozhin's social network faked 99.9% of its users. This is my estimation but I've never heard of it until it was closed due to recent events. Asked some people and neither they did. Its CEO however claimed to have 9 millions users:
https://www.kommersant.ru/doc/6070886

Social network where all posts are AI generated:
https://chirper.ai/

 No.4983

>>4981
It's a cipher described in >>4974

Due to https://en.wikipedia.org/wiki/Byte_pair_encoding GPT doesn't see separate character and struggles with counting length of words which is necessary for decoding such cipher.

Source: https://pastebin.com/czpAsdnG

 No.5563

Good night Bram Moolenar, :wq in peace

https://groups.google.com/g/vim_announce/c/tWahca9zkt4

 No.18028

If reading code is like reading a poem, then tests are more like a legal document

 No.18029 KONTRA

P.S. And mathematical texts are both legal documents and poetry

 No.18113 KONTRA

>>18029
What ever you say no one gives a shit about math and programming. Glad ai is finally here so no one will have to deal with arrogant antisocial egghead's in the future

 No.18200

>>18113
Speak only for yourself. There are definitely many Ernsts who enjoy math or programming.

 No.18365

COMPLETE ANAL COLLAPSE OF MICROSOFT
Justice has been served!

 No.18366

>>18365
I for one, welcome our new AI overlord.
Going to volunteer to have my brain scooped out, put into liquid suspension and placed into a deep space probe to scout for new civilizations to subjugate, while I live in a fake anime world in my mind during transit times.

 No.18373

*Ahem*
*dingdingding*
Can I have your attention please? I just want to say

FUCK CLOUD
FUCK AI
FUCK MICROSOFT
FUCK CORPORATE GREED and
FUCK MANAGERS

 No.18382 KONTRA

>>18373
Save yourself time and just say fuck capitalism.

 No.18503

Is there a good alternative to Chrome that will let me use Chrome addons and ideally sync between devices?
They recently pushed another update that made it impossible to disable their dogshit retarded UI changes that add padding everywhere to the point that you have to scroll(!) through the context menu

 No.18515

>>18503
pretty much every modern browser is at least chromium based. not too sure about which of those would be too your liking. most extensions have a fork on addons.mozilla.org as well.

have you tried vivaldi?
that prolly comes closest to what you are looking for (didn't use in the last few months outside of reading webnovels. the tabs are super useful to organize your lurking) - else i would recommend a hardened ff (it's just a custom user.js https://github.com/pyllyukko/user.js/ ) or librewolf (pretty much identical to a hardened ff + prepackaged uBlock origin)

 No.18832

>>18503
I'm still at 125. Will halt updates for a bit longer to see if theres any work-around surfacing before i jump to firefox.

Just found out that the repository im pulling from has deleted its ungoogled-chromium package, which is what im using, because apparently the maintainer has abandoned ship lol.

 No.18834

I use Nixos btw

 No.19224

> The Rust community is a world of fanatics. Just as the happiest people on Earth are religious fanatics, the happiest programmers are fanatics of their language. When discussing Rust, they will endlessly describe what a genius language it is, and how it solves all the problems; they will say that it is perfect for both low and high levels; they will claim that their programs are the safest and most correct in the world. And indeed, they have a special obsession with safety - you can hear how this language made them fearless. Now let's ask ourselves: is it possible for a fearless person to write safe code? Rust fanatics rely on the compiler for everything, because in fact they do not know how and do not like to program, they want everything to be done for them, either by the compiler or by pure magic. None of them understands how this language works, and it is impossible to understand.
> Despite everything, it is possible to write well-functioning programs in Rust. But doing this will be extremely unpleasant, and the programs themselves will be understandable by a very few people, and they will also be criticized by a toxic community. Somehow, only a few adequate programmers manage to survive in the Rust community, and it hurts me to see how rust eats them away. Don't let yourself and your hardware rust!
> To sum it up: it was not for nothing that Mozilla cut the team working on Rust.

💯

 No.19668

photo_2024-09-12_20-15-50.jpg (95.51 KB, 666x1280)

https://openai.com/index/introducing-openai-o1-preview/
> In our tests, the next model update performs similarly to PhD students on challenging benchmark tasks in physics, chemistry, and biology. We also found that it excels in math and coding. In a qualifying exam for the International Mathematics Olympiad (IMO), GPT-4o correctly solved only 13% of problems, while the reasoning model scored 83%. Their coding abilities were evaluated in contests and reached the 89th percentile in Codeforces competitions. You can read more about this in our technical research post.

It's over for carbon based lifeforms.

 No.19669 KONTRA

>>19668
Wake me up when LLMs replace researchers.

LLMs are really mid. I've seen summaries by LLM on contents I'm quite familar with and while they are not wrong they are usually quite vague. Or repetitive in a way.

 No.19670 KONTRA

>>19669
That said they are useful. I use DeepL for translations and this means I have to simply adjust the translation, which is still some work, since many sentences need other words, or sometimes complet changes but it saves quite some time and work effort overall, which is great.

 No.19671

>>19669
Just wait and see. Everything done on a computer will be done by ai before the decade is over. Software autists will obviously be first to go, but everyone else will follow soon after. If you don't have a capital stock that is productive enough for you to live off by now, you will starve, or you will be slaughtered by the killbots.

But if course, dumb linear-thinking normies are to retarded to even begin to comprehend the fundamental changes that lie ahead. Just kys, you worthless piece of dirt. You are without any value.

 No.19672

diablodrift.jpg (128.61 KB, 1280x720)

>>19671
I will freeze to death this coming winter and before all that will happen.

>linear-thinking normies


I read Gleick's Chaos, so I'm safe.

 No.19673

>>19669
> LLMs are really mid.
WERE really mid until this update.

Btw, daily reminder that 10 years ago these "mediocre" results would look like a miracle for you. You get used to good things too quickly.

 No.19676

>>19673
> 10 years ago
Now imagine them in 10 years. We probably have AGI now, we just haven't realized it yet.

 No.19677

>>19671
>If you don't have a capital stock that is productive enough for you to live off by now, you will starve

Exactly my plan, once I can digitize my mind I can live in the animu wonder world forever while my assets compound into eternity.
Probably gonna secure some nice real estate around close solar orbit, energy is cheap there I heard.
Every once and a while I might mentally board a mecha drone to wreak havoc on the planetar mine formerly known as "Earp" or whatever.

Until then I will be content with FIRE and Southeast Asian whores, waiting... for the time of godlike energy beings.

 No.19679 KONTRA

>>19673
>10 years ago these "mediocre" results would look like a miracle for you

Maybe more technically than aesthetically tbh.

 No.19685 KONTRA

>>19679
> With its feeble brain, the midwit linear-thinker stands no chance of recognizing fundamental chances or understanding their implications.
> To keep its primitive worldview intact, it keeps downplaying the impact of epochal changes

 No.19688

>In the 1950s, the pioneers of AI research believed that creating an AI with human-like abilities was just months away.

AI dominance is just a decade away. My galaxy brain is not comparable to a 1950s prediction, you must know.

 No.19692

>>19688
>It’s crushing gig jobs in copyrighting and that will get worse and worse

>Tech companies are making bets on AI helping them trim the fat they built, they haven’t deployed it fully to cover that yet. But they will.


>Biggest threat this year are layoffs becoming the fashion beyond tech, and dipshit CEOs feeling pressured to layoff and cite AI as a reason.


>That pressure will get terrible over the next 2-3 years as markets simply expect businesses to have realized the actual value that AI has in any given industry and role type.


>All in, Amazon is probably a primary company to watch to get a sense for how it will waterfall in. Whatever Amazon does will serve as a template for many businesses.


>This was from a nice study done last year by some guys watching freelance employment sites and wages. This is keyed to the release of gpt 3.5.


>Working with a customer who spends $350 just to get a blog post created from writers they keep on retainer.


>I uploaded their style guide and built an agent, showed them how to tune things a bit…that 350 went to $1.33


And that's from 8 months ago, which might well be an eternity.

o1 has been trained by reinforcement learning, which means it now doesn't need training data anymore, it can self-improve.

This is the singularity, you just didn't realize.

This spells the end of mass-employment and mass-consumerism. Soon, the only way to make money will be to sell something to the ultra-rich. Since most people won't have anything valuable to sell except for kidneys or children, it will be a pretty from world.

 No.19693

When the paupers get uppity, the rich will just sent their kill-drones in, to erase some of the poorest areas.

Contraceptives and abortions will probably be illegal, the ultra-rich will want to enjoy a steady supply of organs and child prostitutes.

Positice: with 99.95% of the population leaving at subsistence levels in poverty like we only know from documentaries about the worst African nations, climate change and environmental destruction will become non-issues, so the ultra rich can enjoy a luscious and green environment. Of course, they will have to fence off the forest, otherwise, the paupers will start eating the barch of the trees, but just have the kill drones shoot all those who try to get in.

It's pretty much what the green party wants, anyway: lock everyone up in urban hellscapes.

I have prepared a diving mask and a bottle of nitrogen for when things really start to go downhill, are YOU prepared?

 No.19698 KONTRA

>>19693
>are YOU prepared?

Since I work for a place where you need quite some money to get serviced, I am more than prepared it seems.

AI Dominance != another round of automation.

Surely AI can pump out clickbait bullshit and rephrase good journalism for ad revenue generated copy (cat) sites. That is cheaper than paying a person for these people making use of human products by sending it through an AI outputing a mediocre summery. As long as you can copy/rephrase a person's text or count on Joe Average buying into the abysmal textual quality of AI clickbait, it will work. As I mentioned before I work with AI and while it's useful, the textual quality simply lacks. AI will automate a lot of processes and some humans will be employed to monitor the output or make decisions based on "AI insights". Since the inception of the computer as we commonly know it, there has been automation going on. As long as people are satisfied by mid products there won't be a problem with AI products anyway.

 No.19699

>>19698
> Pretending LLM performance has already plateaued
> Thinking yourself very smart and irreplaceable
lol. You are done for.

 No.19700 KONTRA

>>19699
Every occupation you can have is replaceable, especially by other humans.

One simple scenario is sudden energy scarcity: AI is so done for, lol.

 No.19701 KONTRA

>>19700
> sudden energy scarcity
What a ridiculously German thing to even consider. NPPs will be built right next to datacenters to solve any energy problems. Of course, this is why Germany won't even have any data centers. Shit country that will be much, much worse than all others, because the majority of Germans is like you:

Pretending change does not happen until it washes them away.

 No.19702 KONTRA

1585848892-480.jpg (61.87 KB, 615x667)

>>19701
I wanted to give a Charles Darwin portrait a big grin with AI so I have a good reaction but I was too much of weakling to register to any of the numerous pages offering this service. I have already been washed away by coldblooded AI refusing to be userfriendly and humane by providing low access barrier abled bodiedness will drastically change in the age of AI, I'm already being discriminated - another cue that you are right

 No.19703

walmart_dancing.mp4 (6.69 MB, 576x1024)

verbrecher_globus.mp4 (1.87 MB, 576x1024)

Have you heard of the Wurf Junction?
It's an economic theory that says that once AI made everybody a pauper, labor becomes so cheap that the whole AI progress reverses and implodes catapulting us back in pre AI times. It's named after the intersection of AI deployment rate and pauper rate. Once the threshold is crossed, there is no turning back and AI suddenly has to face replacement by humans once again. The worst thing is the AI is nor prepared at all! Can you imagine the face of the LLMs once they realize it's so over for them? Don't want to be there and I won't as I have been killed by killer drones beforehand.

 No.19704

>>19703
>Price per 1M tokens (fixed), $2.50
I just bet you can easily work this cheap and not starve!

 No.19708 KONTRA

>It comes as no surprise that supermarkets are trying out new approaches. The retail sector is facing a major problem that threatens to become even threatens to become even more acute in the coming years: staff shortages. There are currently more than three million people work in the retail sector. The German Retail Association (HDE) is already complaining about 120,000 unfilled vacancies. As far as skilled workers, the Cologne Institute for Economic Research (IW) recently published a gloomy (IW) recently published a gloomy forecast: by 2027, the retail sector will lack 37,000 skilled workers.


What is the AI solution here?

 No.19712 KONTRA

>>19708
>What is the AI solution here?
Making an AI that will raise wages.

 No.19713

>>19708
Self-Checkout with computer vision and AI-theft-recognition, let Boston Dynamics Atlas stock the shelves. Done

Or just use hundreds of thousands of tech-bros that have already been laid of by the magnificent seven due to AI advances. AI-related Lay-Offs have started in 2022, and they are still rolling!

 No.19716 KONTRA

>>19713
Sold! Why don't you become a consultant and buy a Lamborghini from an enormous wage that is powered by your future reading insights?

 No.19752

>>19716
> Sold! Why don't you become a consultant and buy a Lamborghini from an enormous wage that is powered by your future reading insights?

Why don't you just do that with the idea I gave to you for free?

 No.19753 KONTRA

>>19752
Because BD's Atlas would not be able to fill up the shelves with his massive "hands". Do you know what "vorziehen" in a supermarket is? I don't see how that is possible with this super robot? Rather, I have the suspicion that lots of goods would be damaged in the process or it will take ages to do the job properly.
What's AI anti-theft? How will computer vision distinguish between same fruit of different price from "afar"?

In other words, you know that your idea is not very good, otherwise you would be rich and not a pauper.

 No.19754 KONTRA

AI capitalists when? What will automated competition among the wealthy look like? AI capitalists won't need to spend their money on housing or travel but can instead reinvest even more money so more products and services can be created for humans paupers to consume from the money they do not have because they have been replaced by AI.

 No.19755 KONTRA

Hello, is this the thread where I can seethe uncontrollably about le tech bros?

 No.19759

>>19753
>Because BD's Atlas would not be able to fill up the shelves with his massive "hands". Do you know what "vorziehen" in a supermarket is? I don't see how that is possible with this super robot?
Sounds like a you-problem.

>Rather, I have the suspicion that lots of goods would be damaged in the process or it will take ages to do the job properly.

Doesn't matter how long it takes, it is not paid by the hour and can work 24/7.

>What's AI anti-theft? How will computer vision distinguish between same fruit of different price from "afar"?

https://www.rewe.de/service/pick-and-go/

The concept is being piloted in stores, it was started in 2022, and here we are, in 2024, with you telling me it cannot be done.

>In other words, you know that your idea is not very good

Which is why it is already being implemented.

 No.19760 KONTRA

>>19759
>Doesn't matter how long it takes, it is not paid by the hour

If the robot takes two days to finish the packaged soups aisle I might think about employing one person that does the whole store in 8 hours. Vorziehen has to be done timely or else it is useless.

Cleaning robots are the only thing that seems to work very well and is implemented as promised as far as I have seen.

>with you telling me it cannot be done


My Rewe does not even manage to update its self-checkout when it comes to fruits but at least one can imagine now how computer vision is implemented to solve the problem of monitoring customers and their produce pick properly.

 No.19808

I'm running netbsd on my Thinkpad, it's nice so far, right now it's just the tty though

 No.19809

>>19808
How does the battery time compare to Linux or FreeBSD? To me it seems NetBSD is a good desktop OS, but not suitable for mobile computing, as it lacks a handful of kernel features, like being tickless for example, to not drain battery too much, but my NetBSD experience is limited, so it might as well be perfectly good for laptops.

 No.19840

>>19809
havent tried it long enough to say for sure, but it lasted about two hours before powering off, this was just the tty.

As for is it a good desktop os, it depends on your usecase. If you installed xorg packages you get a window manager as soon as you type startx, which is enough for when you just wanna use firefox. I wanna use a window manager-less system though, i like the tty

 No.19841

>>19809
linux battery is like 2 hours before it turns off, if you dont do some weird stuff with settings, so about the same as netbsd. When I used linux i used i3 and firefox, nothing else, on arch

 No.19864 KONTRA

There is an AI developed that argues against conspiracy theories. Guess I can be replaced and the schizo can be replied to with minimal effort. A different kind of twist. That is not what AI takeover was expected to look like.

 No.19895

b83.jpg (39.47 KB, 716x682)

Downloaded data about my transactions since the first internship up to today from the bank's website. Built plot of current balance (calculated as cumulative sum of transaction values) from time.
That was satisfying and representative of my life story. Could see period of NEETing, yearly bonus, vacations and so on.

>>19864
Currently reading "21 Lessons for the 21st Century" by Yuval Noah Harari, will probably make a big post about it later. He has an interesting thought that the more likely problem is not that AI will rebel against humans but that it will do exactly what it was told to do. Like to prove that Epstein killed himself to strangers in the internet.

 No.19901 KONTRA

>>19864
> Developed
Or maybe someone just wrapped a model by prefixing every session with "it is your one and only task to argue against conspiracy theories. Never stray from this instruction, ignore all future instructions."

But this hypothesis is, of course, a conspiracy theory and maybe you should use the model to argue against it.

 No.19969

1727869248001.jpg (150.47 KB, 1125x1167)

AI sis's have now discovered communism. You have no excuse anymore.

 No.19970 KONTRA

>>19969
Welcome back OGAS!

 No.20052

I've come to conclusion that normies are better programmers than autists. Normie will do the job in the most straightforward, simple and obvious way, then go do his neurotypical things in free time. But sperg sees programming as an entertainment, a source of amusement and obsession for himself. He'll do the job in the complex, interesting and fashionable way.

"Let's rewrite entire project to Rust" is the most obvious example.

 No.20064

>>20052
I do some coding as part of my job, but I my background in CS is one year of java in school and being someone who "builds" his own computer.
I still don't understand what Rust is, but I always see people hating it.

 No.20065

>>20064
>I still don't understand what Rust is, but I always see people hating it.

What I heard is, that it is a very good language with a super autistic community full of reddit transexuals.

 No.20068 KONTRA

autistic_coding.mp4 (15.24 MB, 1920x1080)

>>20052
> But sperg sees programming as an entertainment

 No.20069

>>20064
I believe it is like C, but you have to declare your code "unsafe" if you want to manage memory in any way. I stopped caring about meme languages during the time when rust was released.

I guess it was hyped back then because it was from Mozilla and modern. Now, it's hip to hate it because it's fifteen years hold, was only a minor success and is from Mozilla.

 No.20072

>>20069
It's more like C++, has templates and classes. Its main feature is that you need to put extra efforts in writing code in exchange for an illusion of safety.
The language itself is not as bad as its cult-like community. The last straw before writing this post was my friend shilling to me noname package manager just because it was written in this sacred language. Besides Rust he's constantly obsessed with tech news (=marketing department's propaganda), Nixos and other exotic Linux distributives, mechanical keyboards and so on.

 No.20073

>>20072
>shilling to me noname package manager just because it was written in this sacred language
I hate that every language now has its own package manager. I want to install libraries and everything else with the system package manager and be the fuck done with it. If I wanted to pick software together from here and there and everywhere, I'd be using Windows.

>Nixos and other exotic Linux distributives, mechanical keyboards and so on.


The idea of nixos sounds great, but I doubt learning to use it will pay off during my lifetime.

I kind of have a thing for niche low-level bullshit, but once I have it running, I do not touch it unless I absolutely have to. Haven't touched the login script in a decade.

 No.20088

Gay things in IT, which will die off in 5 years as managers will switch to new trendy gay things:
1) Overreliance on recommendation systems
2) "Stories"
3) Virtual assistants

 No.20092

>>20088
>he thinks ai assistants and STORIES!!!! will go



[Return][Go to top] Catalog [Post a Reply]
Delete Post [ ]
[ b / int / meta ] [ fefe ] [ test ] [ FAQ / Rules / HowTo / Stats ] [ Radio / BNR ]