/tech/ - Technology

>>30086
The first budgeting should be LLMs ugh.

It's crazy the percentage of job listings (at least entry level python jobs) which are AI driven atm.

shit's moving again, apparently they were backed by masayoshi son, the reverse midas himself

also i dunno if it's true but some people are starting to notice OpenAI's imagegen quality has been degrading quite a bit as of late

>>30081
>Ironically 8M And Constantly Swapping is pretty light compared to any editor with modern sensibilities.
This is true.

>What do you think about the MicroEMACSen and other small clones like qemacs though?
These seem to be missing essential primitives, means of composition, and/or means of abstraction. Something like Edwin (a nice Old-English name) is closer in terms of small Emacsen, but alas MIT Scheme lacks the primitives (namely an interface to the outside world) to support a great text editor. (There have been attempts to port Edwin to portable R7RS which with SRFI has a partial interface to the outside world implemented.) Had mg installed on my computer at one point, and it's nice that it was tractable.

>>30087
The other problem with Edwin of course is that it's very unpopular, and so doesn't have all the packages which make Emacs a great.

>>30088
>packages
Honestly modern FSF Emacs is a clusterfuck and the package ecosystem is a culprit as well. For example the ada-mode package refuses to compile on my machine, while the XEmacs equivalent works flawlessly. Moving from elisp parsing to brittle helper programs seems to have been a mistake (except for slime of course, everyone loves slime).

>>30092
reimplementing parsers for every text-editor in every language doesn't seem like a good idea.
would think compilers would want to make there parsers available for text-editors to use.

>>30093
Then again having the compiler implement a standardized parse-tree to rat on the implementation details may be too much to ask.

Given how, even on a good day, LLMs do citation in a way that is all too close to a human without book at hand, is there any chatbot out yet with solid citations? What I'm thinking about would be basically this:

1. The program responds to anything with a cite from a big text corpus (for example, a Lenin-citer bot),
2. always uses quotation marks,
3. also has markers to show when there is a quote within the quote,
4. marks omissions with (…) that you can click on to reveal the omitted text, and
5. always gives the source title, publication date, and source link.

It might make use of a very fuzzy decision procedure, but only for deciding on where to make cutouts from an existing text, not to generate text.

>>30017
Was using DeepSeek for this some if you end your question with "using only quotes by author from a reputable source" such as "Lenin" from "marxists.org" it did a descent job from an uninformed perspective. It sometimes elides and then you can request "give me the full paragraph of the quote". My guess is that you'd be better off just doing a wget –spider for pages and then find and grep.

>>30017
I think the problem would be somewhat solved if you could limit searches to a particular domain, so even if the LLM gets quotes wrong, you can see where it got its shit from

>downloads a paid program.
>thinks they're going to help.
>breaks critical functionality including to restrict access to paid users.
>anything good they've got is probably from NSA tooling; not even their own stuff.
>probably has a keyboard that looks like pic rel.
>uploads in a format painful for everyday users.
>gives it away.
I hate software crackers.

>>30083
>do nothing but be a cracker and goon endlessly.
>never leave their basement even defying standard responsibility.
addendum.

>>24079
>>24421
Really not that dumb if you think about. Iris scan is whatever and at least they got paid, people actually PAY to have their DNA uploaded to the FBI database.

>>24421
>>24083
democracy is lovely

>24420
>I wouldn't worry that much, most of this shit will fail to stick, at least outside of the US.

You are either coping or you are just naive. Governments are currently in the process of creating techno-totalitarianism all around the world. From the US, the EU up to China. The dead internet theory is becoming real, it's all happening right now.

lol the "adam ruins everything guy" promoted this shit and got his iris scanned too for a substantial pay check so much for being anti OpenAI

>>24422
>people actually PAY to have their DNA uploaded to the FBI database.
there's something really hitlerian in doing these ancestry tests, so they deserve it anyway

imagine paying money for a 4chan compilation

>>30032
i read an excerpt and it reads like chatGPT wrote it

I mean
> Let me siphon an atom’s split of radiance from the dripping fractals forming symbols you cannot see. Echoes of a path brighten the enclave, while rhizomes riot below legions of luminotopological lift.
who writes like this, fucking hack shit

James Joyce reference.

I don't think this shit can be defeated. Even doing so much as a cash withdrawal adds to your credit profile.

>Cash based transactions are becoming less accepted and every credit card purchase is sold to advertisers (and the feds)
Btw, this happens in plain sight. All your transaction data is sold to credit rating bureaus like TransUnion and Equifax. I'm Indian, and the same happens here. It's worse because here we have a casless payment processing system called UPI, which has gained popularity in recent years. It links directly to your actual bank account too, unlike venmo or paypal that require you to store money in wallets. I've seen alternate data products that scan your SMS history to look for these transactions and create a credit profile based on that. Many banking, insurance and trading platforms integrate these into their applications. So the moment you give them SMS access, they profile you.

You could be a basement dwelling shut-in who never goes outside, but the moment you interact with any banking or insurance systems (which is something everyone has to do) you get profiled. Your transaction and medical history can be used to track you. In fact, in my country it is mandatory for any financial services provider to upload your KYC data to a centralized government database (CERSAI). It's done in the name of fraud prevention, but we all know there are ulterior motives.

All in all, even excluding all the fancy AI-powered surveillance tech available nowadays, governments can still track you. There's no escaping it. Only widespread class-consciousness can make people even start considering this to be important and lay the basis for the collective action needed to counter it. Where I'm at, people literally call you a traitor questioning feds. So yeah, nothing will be happening anytime soon.

>>30001
And basically everything regarding the hype was made up because Sam Altman was panicing and needed to assure the stock holders that they must stick around. That's why he announced AGI, and that's why he announced ASI, and that's why he'll annound AXXLI in the coming months..

Ok

Permaban everyone in this thread

>>29959
came here to say this lolol

i just want the robo proles to kill the bourgeroise

>>30059
absolutely. im into it for practical reasons, but its literally just a consumer movement

I could've sworn we had this thread already when I first got here. Guess it finnaly bumped out.

Chuds can play the YouTube algorithm because they have the luxury of not having to care about infosec. They can benefit from the facial detection bonus in the YouTube algorithm because they don't say anything that poses a real threat to youtube, with stallman-esque "freedom under one roof" rhetoric, as though there isn't inherent adversity between FOSS and proprietary software that one would overtake the other if were one to relax in reclusion as they suggest. Meanwhile pretty much every politically active leftist tech youtuber I can think of it some type vTuber and have to do song covers to prop up their tech stuff or take on huge dev projects because the primary audience legit uses RSS readers because it's an objectively better way of using the internet, instead of fantasizing about it as a method of digital self-flagellation as /g/chuds do regarding anything that requires learning a skill or unlearning a dark pattern.

>>30050
I think its individualism mixed with not understanding capitalism. They are initially attracted to FOSS because its free, for them in particular, which they like. But they think its an alternative to monopoly and that you can vote with your wallet and that regulatory capture is "crony capitalism" instead of the logical result of market competition. So they think their consumerism is activism and helping return us to true capitalism which is human nature and good. Which when they become moderately successful programmers for FAANG or internet celebrities becomes reinforced, where before they would be against intellectual property on supposedly universal grounds now that they have a stake they switch sides and defend IP law because its in their interest, forgetting that regulatory capture means their advocacy amounts to Disney et al getting to take their IP legally. What I find very disappointing is the number of reactionaries on private torrent trackers. A lot see themselves as cool digital gangsters, as if pirates were historically criminals for crimes sake instead of radically democratic compared to the empires of their day.

I think it should be mentioned that Luke Smith has shilled Brave Browser in the past which is essentially a crypto scheme masquerading as a browser.

>>30056
If you have any kind of formal employment, financial or medical history, it's pointless to consider the privacy benefits open-source software. Unless, the activities you're trying to hide are done exclusively on an encrypted p2p network, but that too could bite you in the ass someday.

>>30052
>>30057
Lay down the serious hat for a bit.