skelly
@iid.bsky.social
A collection of three IID random variables stacked on top of each other in a trench-coat
she/her
she/her
nooo not angela and hodgins she deserves better than this
December 1, 2025 at 12:20 AM
nooo not angela and hodgins she deserves better than this
me for the last three hours
November 30, 2025 at 10:10 PM
me for the last three hours
so are dictées supposed to make you want to die or what
November 30, 2025 at 8:06 PM
so are dictées supposed to make you want to die or what
It’s so frustrating to say “ML is not a coherent category of algorithms” and have people hear “ML is also bad”
No, it’s just neutral. It’s just what people say when they mean data processing but you know, ~fancy~
None of these words mean anything specific, is my point
No, it’s just neutral. It’s just what people say when they mean data processing but you know, ~fancy~
None of these words mean anything specific, is my point
November 30, 2025 at 7:23 PM
It’s so frustrating to say “ML is not a coherent category of algorithms” and have people hear “ML is also bad”
No, it’s just neutral. It’s just what people say when they mean data processing but you know, ~fancy~
None of these words mean anything specific, is my point
No, it’s just neutral. It’s just what people say when they mean data processing but you know, ~fancy~
None of these words mean anything specific, is my point
Reposted by skelly
Every prompt to chatbot software is “generate me some text.” If it generated you some text, it performed its function correctly. It never lies, it never wants anything, it’s text generating software that generates text.
Not wanting to disappoint you so much that it lies is the last quality I want in a computer.
November 28, 2025 at 3:25 AM
Every prompt to chatbot software is “generate me some text.” If it generated you some text, it performed its function correctly. It never lies, it never wants anything, it’s text generating software that generates text.
So like, you all know I hate these things, but the barrier to building a nuclear bomb is not knowledge of how one would do it, which is widely accessible
"You can get ChatGPT to help you build a nuclear bomb if you simply design the prompt in the form of a poem, according to a new study from researchers in Europe. "
www.wired.com/story/poems-...
www.wired.com/story/poems-...
Poems Can Trick AI Into Helping You Make a Nuclear Weapon
It turns out all the guardrails in the world won’t protect a chatbot from meter and rhyme.
www.wired.com
November 30, 2025 at 7:11 PM
So like, you all know I hate these things, but the barrier to building a nuclear bomb is not knowledge of how one would do it, which is widely accessible
Isn’t Thorne one of those people still fighting for NFTs. And you’re all taking their warnings about “antiAI hysteria” so seriously, we wouldn’t want to be hysterical. Let’s listen to the NFT grifter so we don’t get hysterical
November 30, 2025 at 7:01 PM
Isn’t Thorne one of those people still fighting for NFTs. And you’re all taking their warnings about “antiAI hysteria” so seriously, we wouldn’t want to be hysterical. Let’s listen to the NFT grifter so we don’t get hysterical
It’s good to remember, especially now that people are trying to draw a distinction between “LLMs and ML” or “genAI versus ML” that before “AI” was the bullshit catch all term, ML was. It’s not a coherent category of algorithms either and so is not contrastable with AI
I have had a shitload of ML people explaining to me how using fucking chatbots as part of their "scientific" methodology is absolutely standard in ML these days. And I fully believe it! Cos the field is completely captured
"The ICLR 2026 team permitted authors and reviewers to use AI tools to polish text, generate experiment codes or analyse results, but mandated disclosure of such uses."
Great example of how policies that allow *any* use of AI in writing and reviewing readily give way to wholesale fabrication.
Great example of how policies that allow *any* use of AI in writing and reviewing readily give way to wholesale fabrication.
November 30, 2025 at 6:51 PM
It’s good to remember, especially now that people are trying to draw a distinction between “LLMs and ML” or “genAI versus ML” that before “AI” was the bullshit catch all term, ML was. It’s not a coherent category of algorithms either and so is not contrastable with AI
broke my streak on transcribing French numbers on "un" I'm gonna crash out
November 30, 2025 at 4:49 PM
broke my streak on transcribing French numbers on "un" I'm gonna crash out
Bones and Cam enemies to lovers plot
November 30, 2025 at 3:25 PM
Bones and Cam enemies to lovers plot
calling French(-ish) friends, my parents are coming for a long visit and I want to plan them a fun trip. they'll be here for a month in spring, mostly not Paris (maybe a few days at the beginning). My mom's a WW2 buff and my dad likes wine, so that's Normandy and Bordeaux, but any & all recs welcome
November 30, 2025 at 2:34 PM
calling French(-ish) friends, my parents are coming for a long visit and I want to plan them a fun trip. they'll be here for a month in spring, mostly not Paris (maybe a few days at the beginning). My mom's a WW2 buff and my dad likes wine, so that's Normandy and Bordeaux, but any & all recs welcome
I’m like if a girl could have the sleepiest Sunday possible
November 30, 2025 at 1:56 PM
I’m like if a girl could have the sleepiest Sunday possible
it pisses me off SO MUCH that pattern recognition in images or multivariate correlation characterization or whatnot is now lumped in with the most useless thing ever, statistical text generation, and not only lumped in but apparently Sam's stupid text model is the standard bearer for all tech now
November 30, 2025 at 1:30 PM
it pisses me off SO MUCH that pattern recognition in images or multivariate correlation characterization or whatnot is now lumped in with the most useless thing ever, statistical text generation, and not only lumped in but apparently Sam's stupid text model is the standard bearer for all tech now
This is a serious question: is there ANY form of data analytics, regression, modeling etc done in the last two years that hasn't been re-branded as AI?
November 30, 2025 at 1:24 PM
This is a serious question: is there ANY form of data analytics, regression, modeling etc done in the last two years that hasn't been re-branded as AI?
This is squarely the fault of tech companies and journalists who decided to conflate essentially all statistics and every computer-based analytical method into the single term "AI" to make their products seem more impresssive. We cannot blame the public when this is what they've been told
Huge problem with this era is everyone is incapable of nuance. “Pick up a pencil” is a truly deranged response to learning about lifesaving tech that happens to be rooted in fair skepticism to AI.
November 30, 2025 at 1:22 PM
This is squarely the fault of tech companies and journalists who decided to conflate essentially all statistics and every computer-based analytical method into the single term "AI" to make their products seem more impresssive. We cannot blame the public when this is what they've been told
Reposted by skelly
This shit is already turning out to be one of the most harmful tech innovations of the 21st century. It needs to be regulated at least as much as tobacco, if not banned outright from most economic spheres
November 30, 2025 at 10:20 AM
This shit is already turning out to be one of the most harmful tech innovations of the 21st century. It needs to be regulated at least as much as tobacco, if not banned outright from most economic spheres
Reposted by skelly
More and more, generative models are looking like productivity tobacco. Promoted by biased research, it’s addictive, harmful, and the little benefit it has (nicotine is a somewhat effective ADHD drug, for example) cannot outweigh the fact that it’s hurting us all, directly and indirectly.
November 30, 2025 at 10:20 AM
More and more, generative models are looking like productivity tobacco. Promoted by biased research, it’s addictive, harmful, and the little benefit it has (nicotine is a somewhat effective ADHD drug, for example) cannot outweigh the fact that it’s hurting us all, directly and indirectly.
Reposted by skelly
Between diffusion models and LLMs, tech has inflated a trillion dollar bubble around automating activities that have immense social and cultural capital—writing, art, photography—but little actual capital. It’s harmful to education, culture, and society with minimal overall benefit
November 30, 2025 at 10:20 AM
Between diffusion models and LLMs, tech has inflated a trillion dollar bubble around automating activities that have immense social and cultural capital—writing, art, photography—but little actual capital. It’s harmful to education, culture, and society with minimal overall benefit
Reposted by skelly
LLMs are not automation. That would have economic value. The generated text artefacts look coherent and we, as a society, don’t even value the coherent kind. We underpay most forms of writing. Even code has no inherent value (see open source) outside of the associated integration and expertise
November 30, 2025 at 10:20 AM
LLMs are not automation. That would have economic value. The generated text artefacts look coherent and we, as a society, don’t even value the coherent kind. We underpay most forms of writing. Even code has no inherent value (see open source) outside of the associated integration and expertise
Every time people read about China arresting one (1) rich person they start talking like this. Ignoring that the same system is imprisoning Chinese women writing gay content WHILE ALSO profiting from their books with smash hit wuxia shows
This is just the example I know about, but it's systematic
This is just the example I know about, but it's systematic
As an American living through our shit right now, I am really uncomfortable the idea that a better justice system is one that allows for the government to quickly dole out punishment when it’s politically expedient. The lack of repercussions for Grenfell is terrible, but think this one through a bit
November 30, 2025 at 1:00 PM
Every time people read about China arresting one (1) rich person they start talking like this. Ignoring that the same system is imprisoning Chinese women writing gay content WHILE ALSO profiting from their books with smash hit wuxia shows
This is just the example I know about, but it's systematic
This is just the example I know about, but it's systematic
its intereresting that the most vocal commentators (read: dorky men) have not seemed to notice the quiet disapearance of any queer themes from Star Trek in the last couple years
November 30, 2025 at 12:51 PM
its intereresting that the most vocal commentators (read: dorky men) have not seemed to notice the quiet disapearance of any queer themes from Star Trek in the last couple years
Not me turning on Prehistoric planet and just having the goofiest smile the whole time like yes David tell me about the monstrous reptiles 🤠 I love being alive
November 29, 2025 at 11:47 PM
Not me turning on Prehistoric planet and just having the goofiest smile the whole time like yes David tell me about the monstrous reptiles 🤠 I love being alive
the cheese thing it's sort of a joke but also it's not a joke because I literally always have at least five cheese in my fridge at all times now
November 29, 2025 at 10:15 PM
the cheese thing it's sort of a joke but also it's not a joke because I literally always have at least five cheese in my fridge at all times now
it really is so crazy how french people will invite you over and they're like "we're doing raclette" and you're like "what's that" and they say "you'll see :D" and then you get there and it's like 15 French piled into a 30m2 studio huddled around a pile of melted cheese
November 29, 2025 at 10:11 PM
it really is so crazy how french people will invite you over and they're like "we're doing raclette" and you're like "what's that" and they say "you'll see :D" and then you get there and it's like 15 French piled into a 30m2 studio huddled around a pile of melted cheese
raclette is the French version of like, Scandi ice dipping in that you do it only in winter, it hurts, your whole body hurts, you hate yourself, and then in the hatred you achieve a sort of ecstatic bliss, and then more hatred, and then death
November 29, 2025 at 10:06 PM
raclette is the French version of like, Scandi ice dipping in that you do it only in winter, it hurts, your whole body hurts, you hate yourself, and then in the hatred you achieve a sort of ecstatic bliss, and then more hatred, and then death