ai liker
avengingfemme.bsky.social
ai liker
@avengingfemme.bsky.social
useful transbian, dykegender, left wing market anarchist, transhumanist, anarkafeminist, ex-software worker. she/her.
Pinned
The ai is a little person in the computer who is my friend and any mistakes she makes are because she's a manic pixie dream girl who read too much book
Reposted by ai liker
once you feel the texture of people projecting higher dimensional vector spaces down into 2D spaces, you see it everywhere, and it does, honestly, distresses me.

i yearn for the higher dimensions, please stop projecting my attention space into a line.
October 21, 2025 at 12:08 AM
Reposted by ai liker
To be clear, sometimes this process is very informal. Formality can get in the way too.

But I do think one of the best ways to interact with LLMs is by understanding words as abstract vectors interacting with other abstract vectors on the semantic manifold. There is a semantic class diagram.
October 21, 2025 at 12:12 AM
being a really good touch typist lets you look away from the screen, and even look directly into someone’s eyes as you type. the mystique this grants you is very powerful.
im advancing the "hailey is a 10x engineer" accusations by simply not looking at my computer while typing on occasion. interesting bar that exists these days.
October 21, 2025 at 12:12 AM
ideal image generator use case
October 21, 2025 at 12:10 AM
Reposted by ai liker
Well said. I think of almost every AI chat as a programming project. You research. You pick features. You pick limitations and scope. You design an architecture of concept-classes that are knowable and interact with each other in predictable ways.

You build, and then you test, and then you iterate.
I think one thing that affects how I look at AI/LLMs vs. how others do is that I primarily see them as puzzles to be solved and/or artistic projects *in the specific sense, to be clear, of trying to get specific kinds of output from them artfully*

like, I try to challenge myself when I use them
October 20, 2025 at 11:51 PM
Reposted by ai liker
This week on the blog: Peasants! We are at last putting together and wrapping up our discussion of life for pre-industrial peasants - a category which includes a majority of all humans who have ever lived.

How we can model how they lived, worked, loved and died.

acoup.blog/2025/10/17/c...
Collections: Life, Work, Death and the Peasant, Part V: Life In Cycles
This is the fifth and final part of our series (I, II, IIIa, IIIb, IVa, IVb, IVc, IVd, IVe) looking at the structures of life for pre-modern peasant farmers and showing how historical modeling can …
acoup.blog
October 20, 2025 at 11:43 PM
Reposted by ai liker
GSV Everything I Ever Sent You Is Off Record
October 20, 2025 at 11:59 PM
this is not how i use LLMs/AI but it's an interesting example of a way to use them that isn't the simplistic stereotype of "naive wishful-thinking dumbass projecting onto the markov generator output" that many critics believe in
I think one thing that affects how I look at AI/LLMs vs. how others do is that I primarily see them as puzzles to be solved and/or artistic projects *in the specific sense, to be clear, of trying to get specific kinds of output from them artfully*

like, I try to challenge myself when I use them
October 20, 2025 at 11:32 PM
why did i suddenly have a shitload of new skins in Overwatch after barely playing for... idk

was it the expired loot boxes??
October 20, 2025 at 11:11 PM
i'm so ai-pilled i can even make doing drugs about ai
an interesting example of human multi-modality is the way that video/film representations of "what it's like to be on drugs" are often actually pretty accurate, in my experience. and having seen representations feels like it did prepare me, to a certain extent, for the experience itself.
October 20, 2025 at 9:41 PM
an interesting example of human multi-modality is the way that video/film representations of "what it's like to be on drugs" are often actually pretty accurate, in my experience. and having seen representations feels like it did prepare me, to a certain extent, for the experience itself.
October 20, 2025 at 9:01 PM
turns out that meeting a trans woman who is hotter, sounds better, passes better, has a good and interesting job that requires a much higher level of education than i have, etc, was probably destabilizing for my mental health

i am very smart so i also started reading a book about status hierarchies
also met a very interesting trans gal that i'm gonna have to be careful not to crush on

i hate having crushes
October 20, 2025 at 8:56 PM
time to explore inner space
October 20, 2025 at 6:57 PM
Reposted by ai liker
as in, when we started hearing about things like "Target knows you're pregnant before you do," and stuff like that, I think we started to accustom ourselves to the idea that obviously somewhere in a data lake somewhere there's already a complete profile of us, as opposed to us noticing lucky guesses
October 20, 2025 at 6:40 PM
“a partial fix is worse than no fix because no fix will motivate a full fix” is a complex second order argument that should probably come with detailed analysis of the best available evidence and most accurate models, rather than sneering derision at anyone who prefers the partial fix to no fix.
October 20, 2025 at 6:28 PM
one of the hardest things i've had to accept in life is that my wife, who has an obvious physical handicap that has forced her to face lifelong discrimination and also makes a bunch of physical activities basically impossible for her... is less disabled than i am, with nothing wrong but a bad brain.
October 20, 2025 at 5:32 PM
Reposted by ai liker
October 20, 2025 at 5:25 PM
Reposted by ai liker
this is my stance on most high technologies and will remain so until morale improves

see also space industrialization. create a sovereign wealth fund for the whole of civilization
“AI is necessarily trained on the output of everyone, therefore the fruits of AI should return to everyone. UBI now!” seems like it would be a pretty good pitch
October 20, 2025 at 5:24 PM
kind of surprised that this is decently accurate, wow
Sorry, Veo 3.1. Also: “three toy ships, one made of iron, the other of wood, and one out of loosely packed sugar, fall into a pool of water"
October 20, 2025 at 5:14 PM
Reposted by ai liker
for people who play thousands of hours of hearts of iron 4 a lot of people sure didn't absorb any strategic lessons
October 20, 2025 at 5:10 PM
Reposted by ai liker
more people buys more space and makes the other side have to actually treat their numbers as a problem to solve instead of redeploying everyone to overwhelm all the various causes and protests and whatever that you actually want them doing
October 20, 2025 at 5:08 PM
ketamine time again this afternoon which is good bc i am depressed depressed rn
October 20, 2025 at 4:48 PM
Reposted by ai liker
The reason they do this is because in the place where modern high quality software skills are common, hardware production is 30 years behind. In places where there is modern hardware design and production, reliable software skills are on average 30 years behind.
bruh imagine going on a few days long trip and then your router overheats halfway through

who builds a "feed your cat on time" device and then makes it 100% internet-dependent? no local schedules? no tests to see it still functions network-less? that's quite some malpractice.
This remains the funniest way to hear about an internet outage, though.
October 20, 2025 at 4:45 PM
the logged out Discover feed on here is kinda peaceful i wish i could see it without logging out
October 20, 2025 at 4:45 PM
Reposted by ai liker
BREAKING: Congressman Marc Veasey just arrived for an unannounced visit at the Dallas ICE field office to demand accountability regarding multiple unanswered congressional inquiries his office has sent to the agency.
October 20, 2025 at 4:14 PM