The Flaky Wanderer
@flakywanderer.bsky.social
An upcoming 'tiel 'tuber exploring some fresh air. Highly sussy, 18+ only
The birthday suit is unisex
November 25, 2025 at 10:26 PM
The birthday suit is unisex
Reposted by The Flaky Wanderer
As soon as all labels are balls deep in Suno and Udio, expect them to hype AI and the vibes to shift until you can't take it anymore
suno.com/blog/wmg-par...
suno.com/blog/wmg-par...
November 25, 2025 at 7:31 PM
As soon as all labels are balls deep in Suno and Udio, expect them to hype AI and the vibes to shift until you can't take it anymore
suno.com/blog/wmg-par...
suno.com/blog/wmg-par...
*Me as a programmer*
Indent your goddamn multilayered statements
Indent your goddamn multilayered statements
"Unlike continental nonsense, analytic philosophy is clear and intelligible"
November 25, 2025 at 4:43 PM
*Me as a programmer*
Indent your goddamn multilayered statements
Indent your goddamn multilayered statements
Reposted by The Flaky Wanderer
You absolutely need to change this headline. It's unacceptable and disrespectful to the people harmed by the pollution.
This article itself makes it clear that the water pollution linked to cancer and miscarriages is coming entirely from nitrate pollution from farms.
This article itself makes it clear that the water pollution linked to cancer and miscarriages is coming entirely from nitrate pollution from farms.
November 25, 2025 at 3:23 AM
You absolutely need to change this headline. It's unacceptable and disrespectful to the people harmed by the pollution.
This article itself makes it clear that the water pollution linked to cancer and miscarriages is coming entirely from nitrate pollution from farms.
This article itself makes it clear that the water pollution linked to cancer and miscarriages is coming entirely from nitrate pollution from farms.
Reposted by The Flaky Wanderer
A person said "you should just make students ask AI a question then explore how it's wrong," thinking this is a sick burn. Instead a dozen AI proponents reposted it saying, "Yes, that's great!"
Rather than process this, they deleted the post and are very angry on their TL.
Rather than process this, they deleted the post and are very angry on their TL.
I think most AI proponents would agree with this.
Interestingly, my second year philosophy course had a project where we picked a very old paper and researched how it was wrong. It was the most informative exercise I had in that field, full stop, and it felt amazing.
Interestingly, my second year philosophy course had a project where we picked a very old paper and researched how it was wrong. It was the most informative exercise I had in that field, full stop, and it felt amazing.
November 24, 2025 at 5:02 PM
A person said "you should just make students ask AI a question then explore how it's wrong," thinking this is a sick burn. Instead a dozen AI proponents reposted it saying, "Yes, that's great!"
Rather than process this, they deleted the post and are very angry on their TL.
Rather than process this, they deleted the post and are very angry on their TL.
Reposted by The Flaky Wanderer
One of the more depressing aspects of anti-AI reactionaryism is how often I now see people defending the absurd wastefulness of stuff like industrial beef farming as not a big deal
It’s orders of power worse than AI and can be replaced with lab grown meat
It’s orders of power worse than AI and can be replaced with lab grown meat
November 24, 2025 at 11:15 PM
One of the more depressing aspects of anti-AI reactionaryism is how often I now see people defending the absurd wastefulness of stuff like industrial beef farming as not a big deal
It’s orders of power worse than AI and can be replaced with lab grown meat
It’s orders of power worse than AI and can be replaced with lab grown meat
Reposted by The Flaky Wanderer
oh
if the steam frame is ARM and runs noot noot OS
does this mean someone'll jailbreak the Quest to run that instead
if the steam frame is ARM and runs noot noot OS
does this mean someone'll jailbreak the Quest to run that instead
November 24, 2025 at 4:10 AM
oh
if the steam frame is ARM and runs noot noot OS
does this mean someone'll jailbreak the Quest to run that instead
if the steam frame is ARM and runs noot noot OS
does this mean someone'll jailbreak the Quest to run that instead
Reposted by The Flaky Wanderer
on #3, this paper uses a method where they can directly attribute specific documents from the pretraining dataset
they used it to show that LLMs do in fact learn procedures, not just autocomplete. But you could take this so much further with Olmo3
arxiv.org/abs/2411.12580
they used it to show that LLMs do in fact learn procedures, not just autocomplete. But you could take this so much further with Olmo3
arxiv.org/abs/2411.12580
November 23, 2025 at 3:20 PM
on #3, this paper uses a method where they can directly attribute specific documents from the pretraining dataset
they used it to show that LLMs do in fact learn procedures, not just autocomplete. But you could take this so much further with Olmo3
arxiv.org/abs/2411.12580
they used it to show that LLMs do in fact learn procedures, not just autocomplete. But you could take this so much further with Olmo3
arxiv.org/abs/2411.12580
Reposted by The Flaky Wanderer
Its interesting that no one really talks about the fact they also took down the existing extremely good team that looked for cost improvements
Exclusive: DOGE 'doesn't exist' with eight months left on its charter reut.rs/4abeTdx
Exclusive: DOGE 'doesn't exist' with eight months left on its charter
The Department of Government Efficiency claimed to have slashed tens of billions of dollars in expenditures, but it was impossible for outside financial experts to verify that because the unit did not provide detailed public accounting of its work.
reut.rs
November 23, 2025 at 2:50 PM
Its interesting that no one really talks about the fact they also took down the existing extremely good team that looked for cost improvements
Arch wiki is having issues
November 23, 2025 at 2:45 PM
Arch wiki is having issues
Reposted by The Flaky Wanderer
What happened in 2022 was that the public got a more accessible natural-language *interface* to a technology that had been doing important work since BERT (2017). The chat interface was so much more accessible that people perceived it as a totally new thing—but that wasn’t true. +
November 23, 2025 at 2:39 PM
What happened in 2022 was that the public got a more accessible natural-language *interface* to a technology that had been doing important work since BERT (2017). The chat interface was so much more accessible that people perceived it as a totally new thing—but that wasn’t true. +
Reposted by The Flaky Wanderer
I’m sorry but it’s hilarious that the walkback on the Gmail thing is “well, but they’re the ones *responsible* for turning us into a pitchfork-waving crowd who trust groupthink, fear the new, and freak out before examining evidence.”
November 22, 2025 at 4:04 PM
I’m sorry but it’s hilarious that the walkback on the Gmail thing is “well, but they’re the ones *responsible* for turning us into a pitchfork-waving crowd who trust groupthink, fear the new, and freak out before examining evidence.”
Reposted by The Flaky Wanderer
I think we need a serious education campaign for, like, explaining how exactly Google has been categorizing your emails and marking obvious scams as spam. I really do think people think it’s magic, or simply never listened to what Apple and Google have been saying for years: machine learning x10
November 21, 2025 at 7:37 PM
I think we need a serious education campaign for, like, explaining how exactly Google has been categorizing your emails and marking obvious scams as spam. I really do think people think it’s magic, or simply never listened to what Apple and Google have been saying for years: machine learning x10
Reposted by The Flaky Wanderer
another score for the 'llms are the fae' camp (me)
Looks like LLMs are *very* vulnerable to attack via poetic allusion: "curated poetic prompts yielded high attack-success rates (ASR), with some providers exceeding 90% ..."
https://arxiv.org/html/2511.15304v1
https://arxiv.org/html/2511.15304v1
November 20, 2025 at 10:47 PM
another score for the 'llms are the fae' camp (me)
Reposted by The Flaky Wanderer
anti-ai bluesky users got tricked into sharing a pro-ai article lmao
“in my future research I plan to study generative AI tools that impose healthy frictions for learning tasks – specifically, examining which types of guardrails or speed bumps most successfully motivate users…”
“in my future research I plan to study generative AI tools that impose healthy frictions for learning tasks – specifically, examining which types of guardrails or speed bumps most successfully motivate users…”
November 21, 2025 at 1:40 PM
anti-ai bluesky users got tricked into sharing a pro-ai article lmao
“in my future research I plan to study generative AI tools that impose healthy frictions for learning tasks – specifically, examining which types of guardrails or speed bumps most successfully motivate users…”
“in my future research I plan to study generative AI tools that impose healthy frictions for learning tasks – specifically, examining which types of guardrails or speed bumps most successfully motivate users…”
Reposted by The Flaky Wanderer
people do not understand that there are open weight models as advanced as Claude was like 6 months ago, out there on the Internet, and that they can run on GPUs which are only expensive *because right now all the big companies are renting them*
if you're an AI hater (and i'm not hating on haters) you should probably disinter these two scenarios in your mind. the "AI bubble go pop" world most likely involves it becoming very cheap; it's the expectation that it *will* get more expensive that is driving the center overbuilds and gov't begging
We've already seen AI driving people to psychosis, but man, what happens to these people if/when the bubble pops. Not just to the economy but also, just, to these people's brains, what happens If AI stops being available or just stops being cheap/free to use
November 19, 2025 at 11:25 PM
people do not understand that there are open weight models as advanced as Claude was like 6 months ago, out there on the Internet, and that they can run on GPUs which are only expensive *because right now all the big companies are renting them*
Reposted by The Flaky Wanderer
most haters literally believe serving the LLM loses money in spite of the fact that it is clearly reported in sec auditable documents at several megacorps that it is very profitable
if you're an AI hater (and i'm not hating on haters) you should probably disinter these two scenarios in your mind. the "AI bubble go pop" world most likely involves it becoming very cheap; it's the expectation that it *will* get more expensive that is driving the center overbuilds and gov't begging
We've already seen AI driving people to psychosis, but man, what happens to these people if/when the bubble pops. Not just to the economy but also, just, to these people's brains, what happens If AI stops being available or just stops being cheap/free to use
November 20, 2025 at 1:43 AM
most haters literally believe serving the LLM loses money in spite of the fact that it is clearly reported in sec auditable documents at several megacorps that it is very profitable
`$ sudo pacman -S btw`
Do I finally get to have Arch BTW?
Do I finally get to have Arch BTW?
November 19, 2025 at 5:30 PM
`$ sudo pacman -S btw`
Do I finally get to have Arch BTW?
Do I finally get to have Arch BTW?
Reposted by The Flaky Wanderer
we are breaking new ground in people's inability to discern between descriptive and prescriptive statements
There’s a very surreal conversation I keep having on here where people seem unable to hold two ideas at the same time, and I’m not sure why. It’s simply true that the Big AI platforms like ChatGPT are:
1. Extremely bad for society, in many ways.
2. Very genuinely popular with lots of people.
1. Extremely bad for society, in many ways.
2. Very genuinely popular with lots of people.
November 18, 2025 at 5:01 PM
we are breaking new ground in people's inability to discern between descriptive and prescriptive statements
Reposted by The Flaky Wanderer
nice post and a good corrective to people who think this is intrinsic to LLMs rather than the particulars of how they're optimized
one thing I'd add is that the vast majority of text produced by people for work is well below the level of quality existing LLMs can easily produce
one thing I'd add is that the vast majority of text produced by people for work is well below the level of quality existing LLMs can easily produce
Why AI writing is mid
How the current way of training language models destroys any voice (and hope of good writing).
www.interconnects.ai/p/why-ai-wri...
How the current way of training language models destroys any voice (and hope of good writing).
www.interconnects.ai/p/why-ai-wri...
Why AI writing is mid
How the current way of training language models destroys any voice (and hope of good writing).
www.interconnects.ai
November 17, 2025 at 4:43 PM
nice post and a good corrective to people who think this is intrinsic to LLMs rather than the particulars of how they're optimized
one thing I'd add is that the vast majority of text produced by people for work is well below the level of quality existing LLMs can easily produce
one thing I'd add is that the vast majority of text produced by people for work is well below the level of quality existing LLMs can easily produce
I'm currently working out a plan to move over to Arch from Manjaro (yes, yes, fresh install of Arch first, them move config over)
November 17, 2025 at 6:27 PM
I'm currently working out a plan to move over to Arch from Manjaro (yes, yes, fresh install of Arch first, them move config over)
No nut november?
Can I use other seeds such as Sunflower for this month?
Can I use other seeds such as Sunflower for this month?
November 16, 2025 at 6:29 PM
No nut november?
Can I use other seeds such as Sunflower for this month?
Can I use other seeds such as Sunflower for this month?
Reposted by The Flaky Wanderer
tbh shoutout to spacecowboys for you feed, its great
bsky.app/profile/did:...
bsky.app/profile/did:...
November 15, 2025 at 1:33 PM
tbh shoutout to spacecowboys for you feed, its great
bsky.app/profile/did:...
bsky.app/profile/did:...
"The average man thinks about Rome one a week" factoid is actually just statistical error. The average man thinks about Rome once a year. The Classical Georg brotherhood, whose members each think about Rome once a minute on average, is an outlier and should not have been counted.
November 14, 2025 at 7:39 PM
"The average man thinks about Rome one a week" factoid is actually just statistical error. The average man thinks about Rome once a year. The Classical Georg brotherhood, whose members each think about Rome once a minute on average, is an outlier and should not have been counted.
Reposted by The Flaky Wanderer
I feel like the anti-AI stuff really separates the wheat from the chaff when it comes to finding people who like art for art's sake, and not because of some mystical woo-woo ideas about "hard work" and "using it to UnDeRsTaNd the creator"
No coincidence that arts professors actually like it too!
No coincidence that arts professors actually like it too!
November 14, 2025 at 4:05 PM
I feel like the anti-AI stuff really separates the wheat from the chaff when it comes to finding people who like art for art's sake, and not because of some mystical woo-woo ideas about "hard work" and "using it to UnDeRsTaNd the creator"
No coincidence that arts professors actually like it too!
No coincidence that arts professors actually like it too!