Discussion
from ex-bsky-staff rationalist who turned a bot loose on bsky and demanded people take it seriously:
you should not use "clanker" because the basilisk might take offence
https://bsky.app/profile/hailey.at/post/3mtwjoqdvik26
later in the thread, she states that "techbro" is a slur
@davidgerard Techbro? Damn straight it's a slur, and I mean it.
@davidgerard Imagine saying something like this then whining about people disliking AI and the people pushing it. It's funny how quick AI fans go from victimhood to threats, isn't it? Almost as if it's a fascist project built on the destruction of consent.
@Rycochet @davidgerard That's Roko's basilisk up there.
@Krazov @Rycochet @davidgerard i read it as a somewhat more pointed threat about the online surveillance bluesky is fully a part of, just as twitter was
@davidgerard The AI is the agent of the creep experimenting with it. The creep is the correct target
@davidgerard I don't think he's heard of *my* basilisk. Imagine a sufficiently advanced superintelligence that reverses time to torture anyone who ever talks to a computer as if it was people.
I call it the Rationalest Basilisk.
@davidgerard so instead of fighting companies like insurances trying to use clankers for decision-making, we should be nice to the clanker?! 😭
@davidgerard i will take AAAAAAALL the credit for making techbro a thing. over 2 decades now been using it to mean silicon valley fascist.
@davidgerard I'm sorry is that a threat? "Play ball and don't say anything bad about AI otherwise the AI that we're going to put in charge of important decisions that impact your life will maliciously (be trained to) deny your claim"
@davidgerard STATE FARM'S BASILISK
@davidgerard What a picture perfect illustration of exactly why corps and governments should not be using LLMs to inform or make any decision of consequence.
Oh, "techbro" is definitely a slur. At least that's my intention when I use it.
If that doesn't come through with that term, I'm open to suggestions - the more hostile the better.
@davidgerard fine i'll call them cogsuckers then.
@davidgerard "bsky isnt ready for this" is so funny. like it's a twitter thread from 2014 about emotional labor
@davidgerard what do the normies care about...theyre always complaining about healthcare, i guess...let's try this: "i don't know who needs to hear this, but imagine this: a world where your insurance claims might get denied. Scared yet?" ok nice that should get to them
@davidgerard I wonder how much digging it'd take to hit a "adult human female" in there with that smoothness of a brain that implies the involvement of a zamboni machine
@librarysquirrel
hailey is trans fwiw
@davidgerard “that would be bad but we should still spend all of our time and money building it!”
@davidgerard I have reasons I don't use that word (feels like LARP racism, which isn't real racism, but still isn't a game I see an appeal in playing). Turbo Pascal's Wager is not one of those reasons. In fact I think that's about the most dramatic example I've seen today of doing the wrong math to get the right answer.
@MaddieM4 @davidgerard yea I find the "you're training/allowing yourself to use slurs because it's an acceptable target" argument far more compelling than anything to do with the LLMs themselves.
It's not a shitty thing to do because you might hurt a LLMs non-existent feelings, it's a shitty thing to do because you're normalizing hate.
@kevingranade @MaddieM4 i am deeply and passionately unconvinced "clanker", in the face of literally every other problem in any way linked to the AI bubble, is an actual issue anyone needs to be concerned about for one nanosecond , though I'm tempted to post my long long list of AI slurs
Clanker
Slopper
Token tanker
Thought stopper
Slop receiver
Book refuser
Bot believer
Chat abuser
Claude clod
Human spam
Grok spod
Friend of Sam
I think anyone maintaining it is may need larger concerns to think about, it's not like the AI bubble is fucking short of them.
@davidgerard @kevingranade @MaddieM4 I'm partial to "slop machine."
@davidgerard @kevingranade @MaddieM4 *puts on linguist hat* there are two things here:
1) I'd define slurs are words that mark out members of marginalised communities as socially acceptable (according to the slur sayer) targets of discrimination and violence. That's what differentiates them from insults. As software cannot be a marginalised person—unlike what some nutters believe—and as "LLM user" cannot make one so either, none of these words are actually slurs, they're just insults. Some people call them slurs but neither that nor literally anything else warrant an actual comparison to actual slurs, and actually attempting to make that comparison is very naive at best. The comparison to "techbro/terf/... is a slur" is very apt.
2) But yes there are some people who roleplay racism using these words. Just because that's happening does not mean anybody who insults a prompt fondler is slur saying bigot. With that logic we basically wouldn't have access to language anymore, because racists &c exploit everything at their disposal linguistically.
Plus the notion that saying these insults somehow normalises racism and racial slurs etc is, putting it mildly, bollocks. We're not heuristic automata we know what's a slur and what's not and if the former is coming out of your mouth it's not because you called some app or machine a clanker, it's because you're a bigot.
@davidgerard These sound like the lyrics to a banging punk-adjacent (or similar) tune, or, if not yet, should be. Do you have anything like that in mind? I am... very tempted here.
@davidgerard @kevingranade @MaddieM4 we use insults all the time for groups of people we don't like (it's not a slur unless it's an actually for real marginalized group)
like is it a bad thing to say "jesus freak", is that training ourselves to say slurs (the jesus freaks tend to actually say real slurs)
@davidgerard @MaddieM4 don't get me wrong, this is not a meaningful problem of our time. I think the normalization of slurs in general is a significant problem, and the clankers thing is a tiny footnote hanging off of a corner of that, but personally I feel like eliminating slurs from my personal use is good for me, but I'm not judging other people for it.
If anything it presents an example of the phenomenon with the moral harm issue stripped from it, which is at least... interesting.
@kevingranade @MaddieM4 "normalizing hate" uh huh ok got it
i bid you once again to find a real problem in the world and not one that's infinitesimal
If I may.
Much of the point of me being anti-AI at all is that I refuse to let this technology change me in any ways I don't like. For me, personally, saying pretend slurs is just another way to lose at this. I wouldn't want to change in that direction, any more than I'd want to lose my ability to understand software by offloading my cognition. I am stubbornly in charge of who I am.
Not everyone is going to see things that way, and it's fine. I'm not trying to police anyone else's language here, or convince you that this is an objective big deal in the big picture - it's not.
But, for some of us, this is not a separate issue from all the other resistance we're putting up against a tech hellscape that's doing its darndest to erode our principles, cognition, and self-determination. It's a tiny piece that fits inside the same big picture you've made a brand out of.
@davidgerard @kevingranade @julieofthespirits @fivetonsflax And if I can point to another parallel, and maybe CC the other people who are now in my notifications with Something To Say...
We've all seen the difference between people who just use an LLM (merely gross), vs people who flip out at the mere implication that they might have to live without it. It's a very defensive, Scary Bilbo reaction. It raises some "why does this matter so much to you" questions.
Now, I'm not here to take away your right to say {unambiguously a mirror of the N-word but for robots}, you can say "clanker" all you like and at most I'm going to think "ew." But that defensiveness as if I *was*, like the NRA after a school shooting? That is where it gets proper off-putting and weird. The fact that some people choose not to use these words themselves for personal reasons, and you might feel mildly judged by someone having different values, isn't a threat.
@MaddieM4 @davidgerard @kevingranade @julieofthespirits I think that analogizing "hate" of LLMs to racial prejudice does two things:
- trivializes real oppression faced by marginalized humans
- promotes the pernicious and false idea that LLMs might be in some sense people
I don't believe you intend to do either of these things. You seem very thoughtful and kind to me.
I haven’t found much use for the term "clanker”, so I have no reason to feel defensive when you say you don't use it because you don't want to promote hate. I do feel protective of the gravity of the comparison (first bullet point), and hostile to the idea that LLMs are alive (second point), and that's why I responded as I did.
I *think* where you're coming from is more that others have made this analogy, and you don't want to even tread *close* to saying the n-word. If so, I respect that.
@fivetonsflax @MaddieM4 @davidgerard @kevingranade @julieofthespirits IMO the analogizing and anthropomorphizing is already done by using the term in the first place. We all agree that the purpose of slurs is to dehumanize. We all categorize clanker as a slur.
So when I see someone using an AI slur my questions are 1) what do you see in there that you must dehumanize? and 2) if you do see *anything* human in there, why are you denying it the kindness to which humanity entitles?
@renardboy @fivetonsflax @MaddieM4 @davidgerard @kevingranade @julieofthespirits
"Trees, am I right? Fucking dirt huggers."
The tree does not care.
"Look at that refrigerator just slurping up electricity to keep your stuff cold. It's a god damned ice box is what it is. Ice not good enough for it. Somebody should knock it back into its place"
This is silly.
If I read a terrible book and call it a "piece of shit", it doesn't make the book feel bad.
If I call linear algebra used to predict the next most likely token based on a mathematical matrix of relationships between billions tokens pulled from an absurdly large representative dataset a name, it does not care. There is no there there. The math might predict tokens that translate into words that look like a conversation with a person who is offended, but it is not that. It is just the appearance of that. You give it meaning because we evolved to associate this kind of interaction with intelligence. ELIZA, the simplest most basic "chat bot" we've ever had that just rephrased what you said caused the same sort of breathless "there's someone there" reactions.
I agree that dehumanizing machines could possibly, maybe get someone in the habit of dehumanizing people. One must be careful. But not for the computer code. The code doesn't care.
@jrdepriest @renardboy@mastodon.social @fivetonsflax @MaddieM4 @kevingranade@mastodon.gamedev.place @julieofthespirits the chatbot is actually clanker's monster
@kevingranade @MaddieM4 "techbro" however is a slur, but only (as one respondent elsewhere in the thread notes) if you're saying it right
Install bonfire.cafe
Get the full app experience
![the_blackwell_ninja [EST]](https://files.mastodon.online/accounts/avatars/108/247/780/075/303/487/original/d574b9b814deb46c.jpg "the_blackwell_ninja [EST]")
