Lvxferre

When it comes to how people feel about AI translation, there is a definite distinction between utility and craft. Few object to using AI in the same way as a dictionary, to discern meaning. But translators, of course, do much more than that. As Dawson puts it: “These writers are artists in their own right.”

That’s basically my experience.

LLMs are useful for translation in three situations:

• declension/conjugation table - faster than checking a dictionary
• listing potential translations for a word or expression
• a second row of spell/grammar-proofing, just to catch issues that you didn’t

Past that, LLM-based translations are a sea of slop: they screw up with the tone and style, add stuff not present in the original, repeat sentences, remove critical bits, pick unsuitable synonyms, so goes on. All the bloody time.

And if you’re handling dialogue, they will fuck it up even in shorter excerpts, by making all characters sound the same.

And sadly, my Twitter/𝕏 thread with the company in private message is going nowhere. 😿

Sometimes “friendly” “reminding” a company about the relevant laws does wonders, making them return such “display” of “attentiveness” in a more timely manner. (Translation: they reply faster if you threaten them with the specific law.)

In this case the Ley Federal de Protección al Consumidor, article 17, second paragraph got you covered:

El consumidor podrá exigir directamente a proveedores específicos y a empresas que utilicen información sobre consumidores con fines mercadotécnicos o publicitarios, no ser molestado en su domicilio, lugar de trabajo, dirección electrónica o por cualquier otro medio, para ofrecerle bienes, productos o servicios, y que no le envíen publicidad. Asimismo, el consumidor podrá exigir en todo momento a proveedores y a empresas que utilicen información sobre consumidores con fines mercadotécnicos o publicitarios, que la información relativa a él mismo no sea cedida o transmitida a terceros, salvo que dicha cesión o transmisión sea determinada por una autoridad judicial

Yup, I got that you don’t mean that everyone is a bot there. I just don’t think that there aren’t so many of them as you’re saying; it’s certainly not as much as half the users, or even the activity (bots tend to be more active than actual users).

They’re still wrecking damage on the place though. Eventually they’ll reach a plateau in proportion, but their numbers will go down, alongside the actual users.

Bots are parasites: they only thrive if the host population is large enough to maintain them. Once the hosts are gone, the parasites are gone too.

In other words: botters only bot a platform when they expect human beings to see and interact with the output of their bots. As such they can never become the majority: once they do, botting there becomes pointless.

That applies even to repost bots - you could have other bots upvoting the repost, but you won’t do it unless you can sell the account to an advertiser, and the advertiser will only buy it if they can “reach” an “audience” (i.e. spam humans).

Shorthand for third language [English] speaker. I mean that I’m prone to switch a few words here and there, due to other languages interfering inside my head.

This sounds familiar, almost as if history could perhaps, maybe, just possibly… repeat itself? Nah! (says spez)

Digg, right? Yeah. Perhaps spez even knew that it would repeat, but was smart (and shitty) enough to jump off the ship before it happened.

I’m not expecting a big exodus, but rather a slow decline in both the number of users and their engagement. With a few peaks here and there that seem to revert the downwards trend, but each peak being smaller than the one before.

They won’t be leaving for the same reason as most people here did, pissed at the IPO-related changes (such as killing 3rd party apps). It’ll be more like “…meh, why would I check Reddit? There’s better stuff elsewhere.” We can already see the decline of the content quality in Reddit now; it’ll get only worse over time.

I think that most will end in Discord. Some in Bluesky, and some will simply touch grass. Conservatives might end in Minitrue “truth social” or crap like that.

Facebook might perhaps absorb some of the former Reddit users. It feels disgusting for the privacy conscious, but for them it’ll be a simple matter of not finding interesting stuff in Reddit.

The same applies to Reddit’s liquid profit - for now, that value extraction still creates a small peak on raw profit, to the point that the bottom line became positive; later on the peak will barely reach the surface; later on, value extraction will be necessary to avoid making the bottom line too negative.

I think that most users there are still human beings, but botting has become a big enough problem that the platform can’t be seen as a place for genuine content any more.

Yup, it is 100% relevant! Selling user data is extremely profitable, specially with a large userbase. However, it lowers the value of the platform - it makes users less eager to genuinely contribute with it (due to privacy concerns, seeing it as a “they’re exploiting me!” matter, etc.). As such the data being generated there becomes less useful, less relevant, and less profitable over time, paradoxically enough.

I fucked it up and switched the terms, sorry. Look for “value extraction” instead; you’ll find multiple references to the concept such as this or Mazzucato’s “The Value of Everything”.

To keep it short: you create value when you produce desirable goods/services for the customers; however, when you extract it, you’re picking the value that was already created (by society, your customers, or even your own business) and turning it into profit. The later is faster but unsustainable, as that value doesn’t pop up from nowhere, so when a business shifts from value creation to value extraction it’ll get some quick cash and then go kaboom.

In Reddit’s case, this value is mostly users willing to generate, curate, and share content with the platform, and other users knowing this:

• someone recommends you a product/brand. The person might be wrong, but you were reasonably sure that they aren’t a corporation astroturfing their own product. Someone else might criticise it instead.
• you hop into your favourite subreddit and, while the content there isn’t the best, it’s still good enough - because the mods gave some fucks about growing their subreddits;
• you discuss some controversial topic. You might get dogpiled, but at least you know that the dogs piling you are human beings, that sometimes might listen to reason; a bot will never;
• et cetera.

All that value was being slowly extracted through the last years, but the changes in 2023/2024 did it the hardest.

As I often mention in other communities, this smells like value exploitation extraction* from a distance. Value exploitation extraction typically generates a peak of profit in the short term, but it makes losses even harsher in the long run.

As such I don’t think that Reddit is getting “bigger”. That profit is like someone who lives in a wooden house, dismantling their own home to sell it as lumber; of course they’ll get some quick cash, but it’s still a bad idea.

In a letter to shareholders, Reddit CEO Steve Huffman attributed the recent increase in users to the platform’s AI-powered translation feature.

Let’s pretend for a moment that we can totally trust Huffman’s claim here. Even human translations often get some issues, as nuances and whatnots are not translated, and this generates petty fights, specially in a younger userbase like Reddit’s; with AI tendency to hallucinate, that gets way worse. And even if that was not an issue, a lot of content is simply irrelevant for people outside a certain regional demographic.

*EDIT REASON: I switched the terms, sorry. (C’mon, I’m L3.)

Kind of. @storksforlegs@beehaw.org is right that journalistic standards prevent too much meddling. Plus commercial news defending interests have a better resource for manipulation - instead of lying, they pick which true pieces of info to release as relevant, and paint them one or another way.

For example. Let’s say that Alice insults Bob, and Bob slaps Alice in return. Someone defending Alice would say that she was the victim of aggression, while someone defending Bob would say that he reacted to Alice’s verbal abuse. Neither is false, but they don’t get the full picture. While LLM/A"I" style bullshit be saying instead “Alice picked a puppy and beat it to death with Bob’s face”.

Any outlet accepting the deal should be immediately put into a list of sites spreading disinformation and potentially harmful content.

Yeah, bullshit machine would be awful for that. The way that it works it’s simply too prone to invent parts that don’t exist, or claim that two pieces are compatible when they aren’t [or vice versa].

All I know is I don’t know how to pick parts and what is compatable with ehat

In addition to the site linked by the other user, you can also websearch “is [part1] compatible with [part2]?” and check the results, they’re often useful.

The site that you’ve linked blocked me for some reason, and cost/benefit in Malta is bound to be different from the one here in LatAm, but I’ve recently built a midrange-ish computer, so might as well list what I bought for reference.

• CPU - Ryzen 7 5700X3D. Good cost/benefit ratio, and rather good performance. I had to buy a third party cooler as the CPU doesn’t come with one, so keep that in mind. I considered the Ryzen 5 5600 for budget reasons, too; it might be an option if you want to make the build cheaper.
• Mobo - Gigabyte B550M Aorus Elite. If coupled with the above you need to Q-Flash update the BIOS, but that was relatively painless. So far it’s working great, can’t complain about it.
• RAM - I went for 2*16GB instead, mostly to future-proof my build. The brand is Apacer Nox, I didn’t find people complaining about it and it had a reasonable price.
• SSD - Adata 480GB.
• PSU - Gamdias Cyclops M1-750B, 750W. Frankly my method to look for a PSU was to look for 700~800W ones in a local forum, with the word “porcaria” (rubbish, shit) alongside it so I could see complains, then I found people actually praising this one.

If I convert my overall costs from reals to euros it was around €500, but keep in mind that I didn’t buy a new HDD or a new GPU. GPUs in special are relatively expensive here, I’m hoping that the prices go down next year.

That’s a great analogy. And a fair point - it got burrowed, but it’s still there.

At least when we deal with individuals using the platform. The platform is still listening to you, and sharing it with advertisers; that’s the whole model behind Meta (WhatsApp) and Snapchat. They’re still hearing you, and want to talk with you (shhh, I’ve heard you bought [product]? Here are some offers for even more [product]!), regardless of what you want.

The whole “one individual talking to another” aspect of the internet of the 00s is gone. It feels more and more like an “everyone is talking to you and hearing you, like it or not”. Facebook is only an example of that - and even if it didn’t enshittify, I find unlikely that it would’ve kept that aspect.

I also wonder if my experiences with Orkut wouldn’t be similar to the ones of the author with FB, if only Google didn’t kill Orkut. (It was a big thing here.)

what can we do?

The link itself offers a good first step: Stallman himself should be encouraged to step down, and if he doesn’t the FSF should remove him from its board.

Furthermore we should be backing up both things and, in their failure, backing up a competing entity.

This should be done in a subtle way, though - without causing unnecessary drama. I know, easier said than done.

A silver lining on everything here is that his saner views are likely to be backed up by other people in the libre software movement.

Agreed.

And that’s extremely shitty because it stains all his views over freedom of speech and libre software, even the sensible ones.