Lvxferre

Jet lag is the hell of a drug. And not a fun one.

Is it? [coherent]

Yes when it comes to the relevant info. The anaphoric references are all over the place; he, her, she, man*, they all refer to the same fossil.

*not quite an anaphoric reference, I know. I’m still treating it as one.

I can only really guess whether they’re talking about one or two subjects here.

It’s clearly one. Dated to be six years old, of unknown sex, nicknamed “Tina”.

Why does it show someone cared for the mother as well?

This does not show lack of coherence. Instead it shows the same as the “is it?” from your comment: assuming that a piece of info is clear by context, when it isn’t. [This happens all the time.]

That said, my guess (I’ll repeat for emphasis: this is a guess): I think that this shows that they cared for the mother because, without doing so, the child would’ve died way, way earlier.

That all reads like bad AI writing to me.

I genuinely don’t think so.

Modern LLMs typically don’t leave sentence fragments like “on the territory of modern Spain. Years ago.” They’re consistent with anaphoric references, even when they don’t make sense in the real world. And they don’t screw up with prepositions, like switching “in” with “on”. All those errors are typically human.

On the other hand, LLMs fail hard on a discursive level. They don’t know the topic (in this case, the fossil). At least this error is not present here.

Based on that I think that a better explanation for why this text is so poorly written is “CBA”. The author couldn’t be arsed to review it. Myself wrote a lot of shit like this when drunk, sleepy, or in a rush.

I’ll go a step further and say that the author likely speaks more than one language, and they were copying this stuff from some site in another language that has grammatical gender. I’m saying this because it explains why the anaphoric references are all over the place.

The article is coherent (it conveys the relevant info without contradicting itself) albeit poorly written. Most likely the result of someone getting really sloppy while writing it, perhaps sleep-deprived. But it doesn’t read like AI stuff, nor a translation - cue to “Cova Negra cave” (lit. “Black Cave cave”).

studies how generative AI can replicate and amplify real-world biases

Emphasis mine. That’s a damn important factor, because the deep “learning” models are prone to make human biases worse.

I’m not sure but I think that this is caused by two things:

1. It’ll spam the typical value unless explicitly asked contrariwise, even if the typical value isn’t that common.
2. It might take co-dependent variables as if they were orthogonal, for the sake of weighting the output.

Impacted nomenclature:

• positron negatron - the antiparticle that annihilates in contact with an electron
• electronegativity electropositivity - property typically associated with nonmetallic elements, specially fluorine and oxygen.
• electropositivity electronegativity - counterpart of the above that nobody cares about
• reduction elevation - half-reaction where a substance retrieves more electrons, thus “elevating” its charge; the opposite of oxidation
• oxidation-reduction oxidation-elevation - the full reaction. Also called “elevation-oxidation”.
• redox elox - acronym for the above.

I’m not currently playing the game (lots to do and, well… it’s Cracktorio, you know), but I’m wondering about the impact of those changes on my typical playstyle. It’ll be probably neutral or positive.

The key here is that I only use the fluid mechanics for short-range transportation, and even then I’m likely to force a priority system through pumps; in the mid- or long-range, I’m using barrels all the time, even for intermediates.

Perhaps those changes will force me to revaluate the role of pipes, that would be a net positive. If they don’t, the changes will be simply neutral.

From HN comments:

I just used Groq / llama-7b to classify 20,000 rows of Google sheets data (Sidebar archive links) that would have taken me way longer… Every one I’ve spot checked right now has been correct, and I might write another checker to scan the results just in case. // Even w/ a 20% failure it’s better than not having the classifications

I classified ~1000 GBA game roms files by using their file names to put each in a category folder. It worked like 90% correctly. Used GPT 3.5 and therefore it didn’t adhere to my provided list of categories but they were mostly not wrong otherwise.

Both are best case scenarios for the usage of LLMs: simple categorisation of stuff where mistakes are not a big deal.

[A] I work at Microsoft, though not in AI. This describes Copilot to a T. The demos are spectacular and get you so excited to go use it, but the reality is so underwhelming.

[B] Copilot isn’t underwhelming, it’s shit. What’s impressive is how Microsoft managed to gut GPT-4 to the point of near-uselessness. It refuses to do work even more than OpenAI models refuse to advise on criminal behavior. In my experience, the only thing it does well is scan documents on corporate SharePoint. For anything else, it’s better to copy-paste to a proper GPT-4 yourself.

[C] lol I can’t help but assume that people who think copilot is shit have no idea what they are doing.

[D] I have it enabled company-wide at enterprise level, so I know what it can and can’t do in day-to-day practice. // Here’s an example: I mentioned PowerPoint in my earlier comment. You know what’s the correct way to use AI to make you PowerPoint slides? A way that works? It’s to not use the O365 Copilot inside PowerPoint, but rather, ask GPT-4o in ChatGPT app to use Python and pandoc to make you a PowerPoint.

A: see, it’s this kind of stuff that makes me mock HN as “Reddit LARPing as h4x0rz”. If a Reddit comment starts out by prefacing the alleged authority of the author over a subject, and then makes a claim, there’s high likelihood that the claim is some obtuse shit. Like this - the problem is not just LLMs, it’s Copilot being extra shite.

B: surprisingly sane comment for HN standards, even offering a way to prove their own claim.

C: yeah of course you assume = make shit up. Specially about things that you cannot reliably know. And while shifting the discussion from “what” is said to “who” says it. Muppet.

Author makes good points but suffers from “i am genius and you are an idiot” syndrome which makes it seem mostly the ranting of an asshole vs a coherent article about the state of AI.

Emphasis mine. It’s like “C” from the quote above, except towards the author of the article. Next~

I didn’t find this article refreshing. If anything, it’s just the same dismissive attitude that’s dominating this forum, where AI is perceived as the new blockchain. An actually refreshing perspective would be one that’s optimistic.

I’m glad to see that I’m not the only one who typically doesn’t bother reading HN comments. This guy doesn’t either - otherwise they’d know that most comments are in the opposite direction, blinded with idiocy/faith/stupidity (my bad, I listed three synonyms for the same thing.)

I’m just going to say it. // The author is an idiot who is using insults as a crutch to make his case.

I’m just going to say it: the author of this comment is an idiot who is using insults as a crutch to make his case.

I’m half-joking by being cheeky with the recursion. (It does highlight the hypocrisy though; the commenter is whining about insults while insulting the author.)

Serious now: if you’re unable to extract the argumentation from the insults, or to understand why the insults are there (it’s a rant dammit), odds are that you’d do a great favour for everyone on the internet by going offline. Forever.

“But LLMs are intellig–” PILEDRIVE TIME!!!

*slow clapping*

I’m actually quite interested in machine learning and generative models, specially LLMs. But… frankly? I wish that I was the one saying everything that the author said, including his dry humour. And more importantly, I think that he is being spot on.

People are selling generative models like they were a magical answer for everything and a bit more. It is not. It is just a bloody tool dammit. Sometimes the best for a job, sometimes helpful, sometimes even harmful. And the output is not trustable, and this is a practical problem because it means that you need to cross-check every bloody iot of the output for potential errors.

I think that I’ll join in and drop my own “angry” rant: I want to piledrive the next muppet who claims that the current models are intelligent.

inb4:

1. “But in the fuchure…” - Vomiting certainty over future events.
2. “Do you have proofs it is not intellijant?” - Inversion of the burden of the proof. Prove me that there’s no dragon orbiting Pluto, or that your mum didn’t get syphilis from sharing a cactus dildo with Hitler.
3. “Ackshyually wut u’re definishun of intellijans?” - If you’re willing to waste my time with the “definitions game”, I hope that you’re fine wasting hours defining what a “dragon” is, while I “conveniently” distort the definition to prevent you from proving the above.
4. “y u a sceptic? I dun unrurrstand” - shifting the focus from the topic to the person voicing it. Even then, let’s bite: what did you expect, F.A.I.TH. (filthy assumptions instead of thinking)? Go join a temple dammit. And don’t forget to do some silly chanting while burning an effigy.
5. “Ackshyually than ppl r not intelljant” - you’re probably an example of that. However that does not address your claim. Sucks to be you.

Based on real discussions. Misspelled for funzies.

OK… here’s some dumb bash shit.

#!/bin/bash
i=0; z=0
while [[ $i -le 1000000000000 ]]; do
	o=$(echo "lvxferre/Hello+Fediverse+$i" | sha256sum)
	if [[ $o =~ ^($z) ]]; then
		echo "$i: $o"
		declare -g z="$z""0"
		fi
	if [[ $i == *000000 ]]; then
		echo "$(expr $i / 1000000)M combinations tried..."
		fi
	i=$[$i+1]
	done

Feel free to use it. Just make sure to change lvxferre/Hello+Fediverse+ to something else.

What it does: it generates the SHA256sum for strings starting with whatever you want, and ending in a number, between 0 and 10¹². Then if it finds one starting with “z” zeroes, it prints it alongside the number; then it looks for strings with an additional zero at the start. Each million tests it’ll also print some output so you know that the script didn’t freeze.

Here’s the content of the OP. Relevant tidbit: it was posted in r/ChatGPT.

I followed these steps, but just so happened to check on my mason jar 3-4 days in and saw tiny carbonation bubbles rapidly rising throughout.

I thought that may just be part of the process but double checked with a Google search on day 7 (when there were no bubbles in the container at all).

Turns out I had just grew a botulism culture and garlic in olive oil specifically is a fairly common way to grow this bio-toxins.

Had I not checked on it 3-4 days in I’d have been none the wiser and would have Darwinned my entire family.

Prompt with care and never trust AI dear people…

Okay… this is a lot like saying “whales are fish, all fish live in the sea, so whales live in the sea”. As in: right conclusion, idiotic reasoning.

No, cold infused garlic oil is not safe; that conclusion is correct. However that’s because you simply don’t bloody know what’s there, it’s like playing a Russian roulette - it might be clean, or it might be tainted.

In other words you can’t simply vomit certainty like “I just grew a botulism culture” from the presence of carbonation bubbles dammit. Plenty healthy fermented food items produce carbonation bubbles, including the beer that I’m drinking now or the sour cabbage on my kitchen counter.

And, when it comes to LLMs, the same (right conclusion, idiotic reasoning) applies. Yeah, the output of any LLM is as trustable as what the village idiot says when he’s drunk; but you need a bigger sample than just one idiotic output to say so dammit. And the answer in this case is technically correct anyway. (You can infuse it. You can eat the result. But you aren’t sure if you can eat it more than once.)

This is solely a pet “theory”. As in, nothing backed up by literature.

What if 600kya we had the shift from paralinguistic communication to true language, and that allowed better transmission of culture?

(For references our earliest reconstructions are for Proto-Afro-Asiatic, around 14kya. As in, almost yesterday.)

Besides everything that the author already said, Fandom is also a cockroach motel from the PoV of the communities using it: it’s trivial to create a new wiki there, but:

• you can’t close it down even with universal agreement of your community
• it has obnoxious forking policies intended to keep the community stuck in Fandom
• the old Fandom wiki surface in search results before anything better that you can pull out, not due to quality but due to aggressive pursuit of SEO cancer from Fandom’s part.

For anyone wanting more info, check the Minecraft Wiki. That wiki migrated out of Fandom, so you can see all the barriers imposed by the roach motel.

Speaking on that. I think that it would be damn great if the Fediverse had deeper integration with self-hosted wikis. Forums like Lemmy are great for discussion, but they suck for long-term storage of information - because eventually the info gets flooded with noise.

Yeah, it’s actually good. People use it even for trivial stuff nowadays; and you don’t need a pix key to send stuff, only to receive it. (And as long as your bank allows you to check the account through an actual computer, you don’t need a cell phone either.)

Perhaps the only flaw is shared with the Asian QR codes - scams are a bit of a problem, you could for example tell someone that the transaction will be a value and generate a code demanding a bigger one. But I feel like that’s less of an issue with the system and more with the customer, given that the system shows you who you’re sending money to, and how much, before confirmation.

I’m not informed on Tikkie and Klarna, besides one being Dutch and another Swedish. How do they work?

Brazil ended with a third system: Pix. It boils down to the following:

• The money receiver sends the payer either a “key” or a QR code.
• The payer opens their bank’s app and use it to either paste the key or scan the QR code.
• The payer defines the value, if the code is not dynamic (more on that later).
• Confirm the transaction. An electronic voucher is emitted.

The “key” in question can be your cell phone number, physical/juridical person registre number, e-mail, or even a random number. You can have up to five of them.

Regarding dynamic codes, it’s also possible to generate a key or QR code that applies to a single transaction. Then the value to be paid is already included.

Frankly the system surprised me. It’s actually good and practical; and that’s coming from someone who’s highly suspicious of anything coming from the federal government, and who hates cell phones. [insert old man screaming at clouds meme]

Both Japanese and your typical West African language stick mostly to CV syllables and word-internal pitch variation. In Japanese said pitch variation is due to the accent, in the typical West African languages due to simple tone systems.

Note however that to generalise African languages like this is silly. It’s a lot like saying that English, Russian, Georgian, and Mandarin “sound Eurasian”. Like, if you squint your eyes ears really hard you can find some statistical similarities (like cramming 9001 consonants per syllable - Mandarin doesn’t do that though), and in some cases you’ll find some actual relation (like English and Russian being related to each other), but for most part you’re fooling yourself.

For reference, check !Xóõ, Amharic and Kimbundu. They don’t really sound like each other, do they? Or like Japanese, although perhaps you can kiiiinda see some resemblance with Kimbundu.

Many Plateau State names can pass for Mandarin names. A former Plateau journalist I used to be fond of (because of the exotic musicality of his name) is called Shok Jok. That name could pass for a Mandarin or Cantonese name.

Cantonese perhaps. But you don’t typically have syllables ending in /k/ in Mandarin.

His lexicostatistical analysis found that less than 30 percent of the similar-sounding words between Plateau State languages and China’s Sino-Tibetan languages share similar meanings. Linguists call these kinds of similarities “accidental evidence.”

That happens quite a bit indeed. Specially often across languages with simpler syllabic structure, by simple statistical chance. But even among others; Zompist has a good text about this, if anyone is interested.

And even when the words do sound similar, and have a similar meaning, sometimes they’re completely unrelated. For example, there’s an Australian language called Mbabaram where the word for dog is… “dog”. English influence? Nope - coincidence.

Do you mind if I address this comment alongside your other reply? Both are directly connected.

I was about to disagree, but that’s actually really interesting. Could you expand on that?

If you want to lie without getting caught, your public submission should have neither the hallucinations nor stylistic issues associated with “made by AI”. To do so, you need to consistently review the output of the generator (LLM, diffusion model, etc.) and manually fix it.

In other words, to lie without getting caught you’re getting rid of what makes the output problematic on first place. The problem was never people using AI to do the “heavy lifting” to increase their productivity by 50%; it was instead people increasing the output by 900%, and submitting ten really shitty pics or paragraphs, that look a lot like someone else’s, instead of a decent and original one. Those are the ones who’d get caught, because they’re doing what you called “dumb” (and I agree) - not proof-reading their output.

Regarding code, from your other comment: note that some Linux and *BSD distributions banned AI submissions, like Gentoo and NetBSD. I believe it to be the same deal as news or art.

Sometimes. Sometimes it’s more accurate than anyone in the village.

So does the village idiot. Or a tarot player. Or a coin toss. And you’d still need to be a fool if your writing relies on the output of those three. Or of a LLM bot.

And it’ll be reliably getting better.

You’re distorting the discussion from “now” to “the future”, and then vomiting certainty on future matters. Both things make me conclude that reading your comment further would be solely a waste of my time.

3. If you lie about it and get caught people will correctly call you a liar, ridicule you, and you lose trust. Trust is essential for content creators, so you’re spelling your doom. And if you find a way to lie without getting caught, you aren’t part of the problem anyway.

For those who don’t want to visit Twitter:

The picture is from an account known for posting product “ideas”, mostly for humour/criticism. And the fact that this “idea” sounds believable shows how shitty YouTube has become.

For writers, that “no AI” is not just the equivalent of “100% organic”; it’s also the equivalent as saying “we don’t let the village idiot to write our texts when he’s drunk”.

Because, even as we shed off all paranoia surrounding A"I", those text generators state things that are wrong, without a single shadow of doubt.