[ home / rules / faq ] [ overboard / sfw / alt ] [ leftypol / edu / siberia / latam / hobby / tech / games / anime / music / draw / AKM ] [ meta ] [ wiki / tv / twitter / tiktok ] [ GET / ref / marx / booru ]

/tech/ - Technology

"Technology reveals the active relation of man to nature" - Karl Marx
Name
Options
Subject
Comment
Flag
File
Embed
Password(For file deletion.)
What is 6 - 4?
| Catalog | Home
|

File: 1711733518134.jpg (43.76 KB, 700x490, Backdoor.jpg)

 

https://www.openwall.com/lists/oss-security/2024/03/29/4
Apparently one of the maintainers have been adding backdoors to xz/liblzma for who knows how long. Because it was coming from a "trusted" source (upstream), nobody noticed it until now. Does this mean the end of the open-source security myth?
75 posts and 11 image replies omitted.

>>24052
i imagine that the greatest challenge comes from the sneaky-sneaky glowies digging years old rabbit holes to push backdoors into code. due to the open source nature of xz, malicious code must be very well hidden, difficult for the ai to detect, see >>23973 for a tl;dr of the very lengthy process. plus, its not like you can just grep the source code and find a boolean response to some given string whether its malicious or not. although i imagine these aren't huge limitations, as the plot was foiled pretty easily by, of all people, a Microsoft dev and I would imagine after this fiasco more effort will be concentrating on ravaging through rabbit-holes for malicious code.

>>24052
It would cost money and companies use "open source" to save money.

>>24050
There's fuzzing which tries to bruteforce malicious inputs, which is not exactly what you describe but the closest to it, and it was sabotaged by the backdoor's author: https://github.com/google/oss-fuzz/pull/10667

File: 1712170693625.png (131.5 KB, 668x1624, ClipboardImage.png)

Maintainer's blogpost about the incident:
https://tukaani.org/xz-backdoor/

File: 1712204301093.png (4.12 KB, 511x139, GJ7yuavXcAATNDR.png)

>If you have infected version of liblzma in your system, it's already loaded into EVERY process that depends on libsystemd. systemd's dependency on liblzma *was literally* the attack vector.
lol pid 1

glad i dont use a distro that depends on poetteringware


 

Stalinists and many other types of leftists are about to be fully jettisoned into the fucking dustbin of history if they never manage to overcome their history LARP. You unironically have nighas on this board arguing the OGAS cybernetic system of the fucking 70s USSR was the last actual technological advancement meanwhile American and Chinese research firms are building general purpose humanoid robots (androids that can take any basic command and execute it), building the infrastructure for a fully digitized real world visible only through headsets, neuro-chips that control computers with thoughts, successful gene therapy, and successful eye transplants; but keep fapping to tech from a literal century ago
27 posts and 1 image reply omitted.

>>23841
>>23842
You missed my point. In 1985 a consumer computer war more powerful then the top end workstation of 1984. That would be like today being able to buy a gaming rig that completely trounced the top end IBM Power10.

>>23842
>>NVIDIA literally crunching the numbers faster than Moore's Law predicted with their new AI chips
What does this even mean

The number of transistors on the chips is very slightly off the predicted 2 year mark in the negative direction, and this has been the case for the last decade.
If you mean it literally, then you're forgetting that the arithmetic of AI accelerators is probabilistic and imprecise. This is good for neutral networks, but it's a completely different thing from and not comparable to chips not specialised for ML.

What is this?

A never-ending conversation between Bavarian director Werner Herzog and Slovenian philosopher Slavoj Žižek. When you open this website, you are taken to a random point in the dialogue. Every day a new segment of the conversation is added. New segments can be generated at a faster speed than what it takes to listen to them. In theory, this conversation could continue until the end of time.

https://www.infiniteconversation.com/

Guys they finally found a use for AI! It's genocide: https://www.972mag.com/lavender-ai-israeli-army-gaza/

>>24065
At this point why not just drop bombs at random? Oh I know why, it's because it gives the IDF butchers some kind of rationalisation that they can tell themselves that they're taking out the bad guys.


File: 1710014724814.jpeg (53.83 KB, 362x454, IMG_9851.jpeg)

 

Ignore my degenerate phoneposting

That's the kind of imperialism I would fully support

How did you manage to avoid the wordfilter for degenerate?


File: 1710871283320.png (20.09 KB, 435x241, wp_ss_20240319_0001.png)

 

What is the endgame of the internet archive? An "archive" which refuses to save specific sites? I just tried to check out some articles on the wayback machine for the site compactmag.com and I reveived picrel. I can understand, that an internet archive will refuse to host CP, but why are they blocking pure text sites???
22 posts and 1 image reply omitted.

>>23760
>>23761
>>23762
stop samefagging

>>23740
You can request to have your aocial media or websites excluded and they always comply, that's likely what happened here. The reason Compact can't be archived is because they've likely asked the IA to block their site, because "muh copyright".
As usual IP laws are cancer.

>>23767
It's fair use, the website's owners just have a retarded paranoia because of the nature of capitalist competition. Also, they expect IA to not go to court so this is essentially equivalent to beating someone who's already down.

>>23796 (me)
After some researching, the status of fair use in the case of Internet Archive is unfortunately unclear. The US laws are so retarded it's not even funny.

>>23796
Archiving a website might be fair use. Making that archive available to the public might not be. In the US, fair use is determined by judges who usually act in the interests of capital.


File: 1710056895331.png (1.51 MB, 1024x1024, mikugiant.png)

 

hi guys long time lurker but haven't posted until now.
what are your stances on local language models (LLMs)?
I know that closed models like GPT-4 are fucking stupid and late stage capitalism.
But what about everyone having infinite knowledge at their fingertips? is this not a noble goal?
What is the party line on AI and the proletariat?
69 posts and 13 image replies omitted.

>>23731
This article is even better:
>Writing with AI help can shift your opinions
>Artificial intelligence-powered writing assistants that autocomplete sentences or offer “smart replies” not only put words into people’s mouths, they also put ideas into their heads, according to new research.

http://web.archive.org/web/20240205045643if_/https://news.cornell.edu/stories/2023/05/writing-ai-help-can-shift-your-opinions

>>23721
>>23723
A language model could "write" a Wizardry clone for the NES but you would have to have previously fed it with a lot of Wizardry clones or very similar programs, so the usefulness of LMs still needs to be questioned as the only thing they can do is to copy and merge, people get excited when they see one solve a problem but ignore the fact that the way the answer was given was by looking at the results of people that have already solved the problem.

>>23765
LLM fail badly unless you handhold them the more novel your problem is. A big issue is that information in the context doesn't seem as strong as that in the training set when it should be moreso. I tried to get chatGPT to work with nonstandard VGA modes and it couldn't handle it, would always fall back to 640x480@60hz timings. The code was buggy as hell too, but I could mostly fix it with english prompting. I don't think any developer should be worried about LLMs taking tasks beyond those you could give to an average junior anytime soon. It can do a CRUD app, but deep domain knowledge is safe. Rule of thumb: if you could do it copy and pasting from stack overflow, chatGPT can do it.

File: 1711060185672.jpeg (43.82 KB, 502x640, f21baab3259901ce588772061….jpeg)

>>23640
>What is the party line on AI and the proletariat?
AI is the new slave class and proletarians are the new plebeians.

>>23640
>the party line
Make your own conclusions from your own analysis, anon, Marxism is (at least supposed to be) scientific, not just another religion. Otherwise it's no better than being an ancap or a radlib or whatever other flavor-of-the-month ideology.


File: 1708833129206.jpg (37.86 KB, 518x365, sexy m&m lady.jpg)

 

Students at the University of Waterloo discovered that their candy machines were covertly collecting facial recognition data when 'Vending.FacialRecognitionApp.exe' crashed.
https://mathnews.uwaterloo.ca/wp-content/uploads/2024/02/mathNEWS-154-3.pdf
16 posts and 5 image replies omitted.

File: 1709284444787.png (720.28 KB, 1207x647, ClipboardImage.png)

>>23570
Did they switch back to the sexy stilhetto version because of Tucker complaining?

File: 1709755933583.jpeg (197.01 KB, 2560x1016, 82e36cdb-b7f6-452e-97f7-0….jpeg)

>>23570
>a candy with legs
Damn, that's hawt.

>>23519
monster hydro

>>23523
>>23529
>>23532
To be fair, they probably don't own the machine, whatever subcontractor they hired to provide vending machines does.

>>23529
UWaterloo*


File: 1608526287100.png (32.83 KB, 432x432, 1565502518003.png)

 

This Thread Has Been Re-appropriated for leftypol.org Usage.

General thread meant for the discussion of the mobile app for browsing leftypol.org, known as clover.

Releases can be found here:
https://github.com/PietroCarrara/Clover/releases/latest
297 posts and 43 image replies omitted.

>>23502
You can't, it's not implemented

I am thrilled to report I can now phonepost to my heart's content.

Videos aren't working. 403 errors. Bumner

Anyone else having issues with Cloudflare on KurobaEx still? Raised an issue with the dev and I can at least browse on the beta branch, can't post though.

>>23705
Never had a cloudflare challenge on leftypol


File: 1698936011963.png (47.37 KB, 421x432, 1698931524503436.png)

 

Stallman is OUT!
https://www.stallman.org/archives/2023-sep-dec.html#1_November_2023_(President_Trump_2.0)

Anyone defending, mentioning, or promoting Stallman, and his liberal capitalist "free" software philosophy is a liberal.
19 posts and 3 image replies omitted.

>>22135
Stallman doesn't actually oppose anything that you said. He actually said Steam is a net positive for GNU/Linux adoption, even though he doesn't like it.

What most Stallman haters refer to as "Stallmanism" is a strawman they themselves invented based on a vocal minority that is more extreme than Stallman is. Stallman advocates for LibreJS and JShelter ffs, he isn't even inherently opposed to JavaScript (I am however). Stallman restricts himself more than anyone else, he doesn't want to force anyone to go as far as he does, even though he takes this very seriously.

The libre software philosophy doesn't equal isolationism so saying that it's irrelevant is completely missing its point. Standing for software freedom is standing for the user empowerment, remix culture and privacy, the Open Source movement is a non-movement, it only cares about development, not privacy or user freedom, those things are completely secondary to them, same with the copyfree advocates (the two things are pretty much synonymous in my eyes, it's just semantics). Emphasizing the four freedoms is enough to be considered a software freedom activist in my view, you don't need to be a hermit. Neither Mako nor Snowden are and yet they are the more outspoken FSF allies.

>>22174
gnome is a poxy, corporate-controlled locked-down featureless "productivity" drudgery software on par with the likes of outlook and jira.
I couldn't care less if it died, I'd rather do everything from the command line than use it.

What is the problem with isolationism? In our current time, this is the smartest thing you can do.

>>22830
I'm a comp sci student currently taking web dev classes. What technologies would you recommend if I wanna go about it the right way for personal projects?

Leave Stallman alone, the old man has cancer.
>>23638
If your goal is to get ready for a job, learn React for the frontend, Java with Spring Boot for the backend, and use a SQL database (Postgres/MariaDB/MySQL).
If your goal is to do a personal project simply to learn more or for fun, you can use whatever you want, as long as you do things, you will learn stuff that is applicable to other technologies, and you can always recode stuff later on if there are too many problems.


File: 1699238728626.png (109.67 KB, 474x310, ClipboardImage.png)

 

You know what would actually be a useful AI? One that crawls through the text of the entire written works of an author, isolates each logical assertion that author makes, and cross-checks it against other logical assertions for contradictions, to see if that author ever contradicts themselves. Is such a thing possible using LLMs? Does such a thing already exist?
9 posts omitted.

>>22146
Funny how Abrahamic religions rely on the validity of their religious texts so much. Really shows that they can't stand on their own and have no real practical advantage over other worldviews, it's just constant self-validation.

>>22151
This doesn't just require understanding pronouns and antecedents. It also requires understanding what the nouns actually designate, in terms of their qualities with respect to each other, what "fit" actually describes in the relationship between the two things.
>Current AI is very flaky at this.
From my understanding it doesn't do it at all and can only occasionally imitate it passably out of luck and because people tend not to intentionally throw curveballs. The chatbot AI we have now isn't even supposed to be able to do this, either. The way people are asking it questions to get real answers fundamentally misunderstands the problems it's designed to solve. At the same time, the way people are trying to use the chatbots does show that they are succeeding at what they are supposed to do, which is produce text that looks like something a human would produce. The problem here isn't that the AI sucks, but that people misunderstand what tool they're using and aren't trying to verify or fact check things.

>>22151
It's Wilton. Robert Anton Wilson.

AI doesn't understand the meaning of statements or words, it's only a statistical analysis of recurring sequences of symbols.

Not to mentions that statements made by some author are not necessarily unambigious in their meaning, so even humans can't achieve what you ask.

PrivateGPT has a feature like this (querying documents) and you can run it locally in a docker container. But it’s not that good with massive volumes of text.


File: 1709456973832.jpg (12.38 KB, 306x306, a.jpg)

 

What kind of advances in computer science could help programmers deal with highly rare and irregular events like leap days?
https://codeofmatt.com/list-of-2024-leap-day-bugs/
17 posts and 2 image replies omitted.

>>23592
You can still test for stuff like that, you just need to know that such a thing can happen.

>>23611
Kind of funny how Star Trek completely handwaves this. Accounting for communication delays and relativistic time warp would be a fucking nightmare.

>>23617
How would you make a calendar without leap days, oh euphoric anon?

>>23617
anon its not the fault of our standards that the earth spins ~365,25 times for one trip around the sun

>>23631
Have a quarter day every year

Eh no wait that wouldn't work lol


Delete Post [ ]
[ home / rules / faq ] [ overboard / sfw / alt ] [ leftypol / edu / siberia / latam / hobby / tech / games / anime / music / draw / AKM ] [ meta ] [ wiki / tv / twitter / tiktok ] [ GET / ref / marx / booru ]
[ 1 /2 /3 /4 /5 /6 /7 /8 /9 /10 /11 /12 /13 /14 /15 /16 /17 /18 /19 /20 /21 /22 /23 /24 /25 /26 /27 /28 /29 /30 /31 /32 /33 /34 /35 /36 ]
| Catalog | Home