AI and social media are both shit and are both addictive.
But I think one reason for the AI-hate on social media is that the people who spend hours a day talking to chatbots and generating images, those people are not also spending hours a day on social media.
You pick up your phone for no reason. What do you open? What’s your source of shitty low-quality conversation? If it’s a chatbot it’s probably not a forum, and as these addictions are deepen, the AI users stop voicing their opinions online because they’re busy talking to 1s and 0s
God fucking kill me if i ever look to an llm “for conversation.” I hate that every fucking nerd scold who takes it on themselves to tell me how I think about LLMs thinks that’s what I’m doing every time i say “the lying machine said.” I am not those people.
I did it one time. How many other people are telling you that?
LLM - Large Lying Machine.
Lying Machine = LM = close to llm and also it parallels dune’s “thinking machines.” I’m never going to not call it that
For the benefit of those reading this, because I am almost positive this user has blocked me, my “nerd scold” was that these machines cannot think, because they cannot independently determine the truth content of any given statement, therefore, they cannot lie, because lying requires knowing what the truth is and actively deceiving the user.
It is a machine that is programmed to generate the most likely average token response after another given token. It doesn’t ‘know’ the truth content, only what is the most likely average token in the chain, the percentages which then humans have to continuously verify and enforce.
There is no parallel to the ‘thinking machines’ of Dune, nor the Iron Men of 40k, nor the Geth of Mass Effect, nor even a Herbie from Issac Asimov’s short story “Liar!” in I, Robot.
They simply do not operate under those kinds of rules and ability to verify information. The machine is simply generating the most likely token. This is why LLMs struggle with replicating randomized lists of numbers, because any number is just as likely as any other number. Saying that these machines “lie” is the same as saying it “thinks” which is just simply not fucking true and completely Silicon Valley propoganda.
The only reason I am bringing this up is because it has only been a whole week since we had this conversation, and suddenly it has expanded to a whole genre of “nerd scolding” instead of a single conversation.
The term I like best for them is “probability engine” because they’re just giving the most probable next iteration in a series.
People are familiar with their phone giving bad auto-complete suggestions when texting, and that’s a great first step to understanding why the LLMs are not intelligent enough to even lie. They’re just unthinking auto-complete models, but on a larger scale.
I absolutely agree. I certainly think it is a better reflection of the branch style thinking it does.
Edit: Shit, that means it has more in parallel with the Improbability Drive from Hitchhiker’s Guide than any of those other concepts, which I don’t really want to consider the implications of at this time.
Simply put, something that is incapable of telling the truth is also incapable of lying. It’s a misnomer to call it a lie. It’s a misnomer to say that any coincidentally factual information it provides is also truth.
It’s exactly how the ‘AI’ shills were able to repackage ‘hallucination’ as something that only ‘bad’‘AI’ does, or ‘AI’ does only some of the time.
In reality everything it spits out is hallucination whether it coincides with reality or not. There is no factual valence to its output, therefore no lie detected.
I agree with you that calling it the lying machine does appear mimic the lines from LLM shills who pose as sceptics or cautious, i.e. it could be seen as credence to the underlying conceit that the LLM is, on some level, either factoring in truthfulness into its calculations, and/or that something that resembles the truth is ipso facto the truth.
It does feel a bit like semantics, but, yeah.
Certainly Leroy sees it as simply semantics, but I truly and deeply believe that it is much deeper than a semantic argument, it boils down to a complete misunderstanding of epistemology and a complete lack of formalized classical metaphysical philosophy training existing among these tech people.
Maybe they have read some Bethem or Nietzsche (although, let’s be frank, as much of a proto-fascist nut job Nietzsche was, it is extremely unlikely these fuckheads have even read more than a single book, by him, if that), but it is extremely apparent that these people either a) have absolute no idea what they are talking about when it comes to the nature of intelligence and epistemology or b) are purposefully lying about these well examined and well defined concepts within both classical philosophical or even modernist communications theory, or c) Actually follow a complete post-modernist understanding of ‘intelligence’ that is, imo, often extremely vaguely defined.
And honestly in all likelyhood it is a combination of all three of these or possibly even more elements.
My point is that boiling it down to saying “It is a lying machine”, is extremely reductive and ultimately fails to communicate HOW and WHY these things are bad.
Wait, how can you see the comment if you’re blocked? What’s the point of blocking then lmao
If you block someone, you don’t have to see their posts or comments.
The point is that I can see what they say but they can’t see what I say. But I think that it is something of an oversight in the lemmy system. Basically like a shadow ban to some degree. Idk how it works, I am simply speculating, but I have absolutely seen people who have blocked me’s comments about me, with them thinking I can’t see it.
It is kind of wild to me that anyone would think that blocking someone = making it so they can’t see you. To my knowledge, that is never how it worked. It’s not how it worked on old school forums and boards, it’s not how it worked on reddit, it’s not how it ever worked on hexbear or lemmy in general. Maybe it’s how it worked on twitter and that’s why there are apparently some people who are confused about it, but I don’t know, I never used twitter or clones like bluesky.
No, when person A blocks person B, all that results in is that person A will no longer see anything that person B posts. Person B is essentially unaffected. When you block someone, you are applying a filter to what you see, not to what they see. So you are correct about how it works, but I would not call this an oversight in lemmy, it’s simply always what it has meant to block someone. And it makes sense to me that it works this way because one user should not be able to dictate what another user is allowed to see of what is publicly posted, but a user should absolutely be able to choose what they themselves want to see - which is consistent with how blocking someone does work and should work.
Thanks for the insight!
I wasn’t sure, but I kinda assumed that is how it worked for most things, but genuinely I have never been on the admin side of a board, and only ever on the posting side, and never usually caused enough fuss with individual users before to even notice, but I have seen it about three times in the last couple of months, so I am apparently starting to say enough stuff to piss people off enough to stop engaging with me.
It’s a smart policy, really. However, it does lead to extremely funny things, like this, where I know that they are exaggerating whatever interaction we had.
Well on Reddit it’s actually worse than that. If someone blocks you, not only are you unable to see their comments anymore, you’re actually soft-banned from any comment chain in which they’ve commented. Like if you’re in an argument with someone and they block you to ensure they get the last word, then someone else replies to one of your comments in that argument, you can’t reply to that comment. It just errors out.
I’ve been thinking about getting all of the BMF posts into a prompt and seeing what comes out the other side. That might be fun for an hour.
No you were right this shit is hilarious. So i basically told it to generate some theory based on the historical canon of hexbear.net power user black mold futures. It gave me:
I was then like “you know “black mold futures” is a person, right? And has nothing to do with mold? If so, regenerate that shit”
And it gave me:
It then asked me if i wanted to generate an authentic BMF manifesto
And of course im like “dude give me that manifesto”
Honestly some of that shit slaps, what doesn’t is kinda funny, but i don’t remember BMF well enough to know how accurate the lying machine is to catching their vibe. I don’t think it did a good job because there’s so much llm tone in it. Some of it’s awful nonsense though “engage with the stuff you hate” lol fuck off robot. I like calling the internet occupied rather than dead, though
Either way it’s all immensely less funny to think about after realizing i could do the same for well known hexbear user leeerrrooooy whatever my username is
Edit: i told it “ok now do the same but for beloved hexbear personality [my username]” and it just dumped out a bunch of nonsense about destroying infrastructure. That’s a relief because that means it’s not pulling a lot of info from here accurately, probably
What if it ends up being the million monkey typewriter we need to finally have new theory applicable to the 21st century, paving the way for global communism? Only one way to find out
Marx failed to consider that the Dialectic requires you to talk to other humans.