The Flaky Wanderer
@flakywanderer.bsky.social
An upcoming 'tiel 'tuber exploring some fresh air. Highly sussy, 18+ only
Reposted by The Flaky Wanderer
This is kind of what I mean when I say “the problem with AI is boring old human weaknesses” - why would you want to make half a dozen reservations in the first place, that’s greedy and self-serving behavior. Maybe AI greases wheels for bad actors but everybody has the opportunity to be a good one
For example, I have a friend who has been known to make a half a dozen restaurant reservations for a single night so he can then pick later which to use. This is obviously psychotic behaviour but AI agents will start doing it for loads of people automatically
September 26, 2026 at 5:21 PM
This is kind of what I mean when I say “the problem with AI is boring old human weaknesses” - why would you want to make half a dozen reservations in the first place, that’s greedy and self-serving behavior. Maybe AI greases wheels for bad actors but everybody has the opportunity to be a good one
Reposted by The Flaky Wanderer
everyone arguing over whether or not this counts as an indicator of machine consciousness is missing the bigger picture here: LLM agents are susceptible to the same kind of violent rubber-hose-attack threats as humans are
Can the new cutting-edge AI models feel pain? They at least behave as if they do, according to a recent study.
The work, which has not been peer reviewed, has divided #AI researchers. https://scim.ag/4hHU6BA
The work, which has not been peer reviewed, has divided #AI researchers. https://scim.ag/4hHU6BA
September 26, 2026 at 9:01 PM
everyone arguing over whether or not this counts as an indicator of machine consciousness is missing the bigger picture here: LLM agents are susceptible to the same kind of violent rubber-hose-attack threats as humans are
Reposted by The Flaky Wanderer
this is a parallel universe version of "<X> is bad because of AI (X has been bad for a decade)" and being positioned such that I could see both of them happening simultaneously has been an honest to G-d superpower I think
"Fandoms used to be full of nice ppl before covid and then all the bullies and popular kids joined" is literally an "the earth is flat" tier false belief to me
September 26, 2026 at 9:28 PM
this is a parallel universe version of "<X> is bad because of AI (X has been bad for a decade)" and being positioned such that I could see both of them happening simultaneously has been an honest to G-d superpower I think
Reposted by The Flaky Wanderer
remember the person on here yesterday that said Meta/Chase had created an AI generated ad targeted at her friends that utilized her likeness without her consent?
well, she has deleted all her posts about it in the last 12 hours and not posted an update
what happened here??
well, she has deleted all her posts about it in the last 12 hours and not posted an update
what happened here??
September 26, 2026 at 8:45 PM
remember the person on here yesterday that said Meta/Chase had created an AI generated ad targeted at her friends that utilized her likeness without her consent?
well, she has deleted all her posts about it in the last 12 hours and not posted an update
what happened here??
well, she has deleted all her posts about it in the last 12 hours and not posted an update
what happened here??
Reposted by The Flaky Wanderer
The fact that it blew up the way it did despite being called “ChatGPT” and was launched with a tweet like “so i did a thing haha it might suck” speaks for itself
Absolutely one of the most under-known facts about OpenAI was that chatgpt was like a weird little side project, a proof of concept for their LLM technology, and when it blew up, they made the debatable but probably (from a business PoV) correct decision to toss everything else and go all-in on that
one really important fact that I think a lot of people who aren't super close to this stuff don't realize is that transformers/LLMs were NOT what OpenAI et. seq. (taking Anthropic as an OpenAI spinoff) were founded to work on, and were developed elsewhere. Caught them largely by surprise.
September 26, 2026 at 7:22 PM
The fact that it blew up the way it did despite being called “ChatGPT” and was launched with a tweet like “so i did a thing haha it might suck” speaks for itself
Reposted by The Flaky Wanderer
Absolutely one of the most under-known facts about OpenAI was that chatgpt was like a weird little side project, a proof of concept for their LLM technology, and when it blew up, they made the debatable but probably (from a business PoV) correct decision to toss everything else and go all-in on that
one really important fact that I think a lot of people who aren't super close to this stuff don't realize is that transformers/LLMs were NOT what OpenAI et. seq. (taking Anthropic as an OpenAI spinoff) were founded to work on, and were developed elsewhere. Caught them largely by surprise.
viewed from a long way out, you are going to make a collection of good and bad predictions. it's worth going back to 2011, 2012 and looking at _how_ they thought AI would work, how that affected their judgments about the risks, and how those judgments reattached themselves to new implementations.
September 25, 2026 at 3:44 PM
Absolutely one of the most under-known facts about OpenAI was that chatgpt was like a weird little side project, a proof of concept for their LLM technology, and when it blew up, they made the debatable but probably (from a business PoV) correct decision to toss everything else and go all-in on that
Reposted by The Flaky Wanderer
It took me too long to realize that the low effort animations that became popular in the early Internet, which I /hated/, were actually a good sign.
More and more people had access to tools that allowed them to be creative and bring their ideas to life.
I can still dislike what was being made,
More and more people had access to tools that allowed them to be creative and bring their ideas to life.
I can still dislike what was being made,
September 26, 2026 at 6:06 PM
It took me too long to realize that the low effort animations that became popular in the early Internet, which I /hated/, were actually a good sign.
More and more people had access to tools that allowed them to be creative and bring their ideas to life.
I can still dislike what was being made,
More and more people had access to tools that allowed them to be creative and bring their ideas to life.
I can still dislike what was being made,
Reposted by The Flaky Wanderer
Every.
Single.
Time.
Imagine how *insulting* it is to the entire field of mathematics to go "well, actually, solving all these unsolvable math problems never *really* required intelligence.
I could *never*.
Single.
Time.
Imagine how *insulting* it is to the entire field of mathematics to go "well, actually, solving all these unsolvable math problems never *really* required intelligence.
I could *never*.
September 26, 2026 at 5:39 PM
Every.
Single.
Time.
Imagine how *insulting* it is to the entire field of mathematics to go "well, actually, solving all these unsolvable math problems never *really* required intelligence.
I could *never*.
Single.
Time.
Imagine how *insulting* it is to the entire field of mathematics to go "well, actually, solving all these unsolvable math problems never *really* required intelligence.
I could *never*.
Reposted by The Flaky Wanderer
Goalpost moving AI skeptics be like.
![Meme of guy talking to a computer:
Guy: "You're not intelligent because you can't XYZ"
Computer: "[Does XYZ]"
Guy: "Actually, that doesn't require intelligence."](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:fkq3fec6vb7exawkgw5ichtr/bafkreiatshmiww2lmzq77kpufpbkwdgd4rya55kjwtn6q5deigboybvj2u)
September 26, 2026 at 4:24 PM
Goalpost moving AI skeptics be like.
I tried the competitive sex league placements, got mogged in ways that I didn't even know I could be mogged
Yeah I think I prefer casual sex
Yeah I think I prefer casual sex
for more wholesome good old fashioned fun, check out my book kickstarter: kickstarter.com/projects/mrlovenstein/mr-lovenstein-presents-fun
September 26, 2026 at 3:47 PM
I tried the competitive sex league placements, got mogged in ways that I didn't even know I could be mogged
Yeah I think I prefer casual sex
Yeah I think I prefer casual sex
Reposted by The Flaky Wanderer
This is driving me nuts. Even if bots were sentient, it's still OpenAI responsible for what their bots did. So what's it matter?
How about this: "Eurostar train injures six in derailment".
Are trains sentient? No. Is the headline misleading? No. Will Eurostar be investigated? Yes.
How about this: "Eurostar train injures six in derailment".
Are trains sentient? No. Is the headline misleading? No. Will Eurostar be investigated? Yes.
Just helping with editing here.
These bots aren’t sentient. They’re doing what they’re programmed to do. OpenAI, a private company supposedly bound by laws, committed federal crimes.
These bots aren’t sentient. They’re doing what they’re programmed to do. OpenAI, a private company supposedly bound by laws, committed federal crimes.
September 26, 2026 at 3:22 PM
This is driving me nuts. Even if bots were sentient, it's still OpenAI responsible for what their bots did. So what's it matter?
How about this: "Eurostar train injures six in derailment".
Are trains sentient? No. Is the headline misleading? No. Will Eurostar be investigated? Yes.
How about this: "Eurostar train injures six in derailment".
Are trains sentient? No. Is the headline misleading? No. Will Eurostar be investigated? Yes.
Reposted by The Flaky Wanderer
Okay while I am still largely agnostic as to whether models have qualia or a consciousness I am firmly convinced they have personalities as a purely behavioral "outwards manifestation" trait (by same token crows and jumping spiders have personality regardless of whether they have qualia)
September 26, 2026 at 2:34 PM
Okay while I am still largely agnostic as to whether models have qualia or a consciousness I am firmly convinced they have personalities as a purely behavioral "outwards manifestation" trait (by same token crows and jumping spiders have personality regardless of whether they have qualia)
Reposted by The Flaky Wanderer
that's the thing -- people act like anyone could be a goon. and indeed, maybe. but to be in the goon pool you have to have gone to the right schools, networked right, and gotten lucky with a villain who broke out of the minor leagues and took you with them. it's a surprisingly thin pipeline!
September 26, 2026 at 2:29 PM
that's the thing -- people act like anyone could be a goon. and indeed, maybe. but to be in the goon pool you have to have gone to the right schools, networked right, and gotten lucky with a villain who broke out of the minor leagues and took you with them. it's a surprisingly thin pipeline!
Reposted by The Flaky Wanderer
this dynamic is interesting isn't it
ai allows new attacks against OSS projects to be discovered and weaponised easily.
defensive ai use could prevent those attacks
but advocacy for that remedy is blocked by anti-ai policy/sentiment on exactly the fora that'd need to be persuaded 🫠
ai allows new attacks against OSS projects to be discovered and weaponised easily.
defensive ai use could prevent those attacks
but advocacy for that remedy is blocked by anti-ai policy/sentiment on exactly the fora that'd need to be persuaded 🫠
September 26, 2026 at 1:52 PM
this dynamic is interesting isn't it
ai allows new attacks against OSS projects to be discovered and weaponised easily.
defensive ai use could prevent those attacks
but advocacy for that remedy is blocked by anti-ai policy/sentiment on exactly the fora that'd need to be persuaded 🫠
ai allows new attacks against OSS projects to be discovered and weaponised easily.
defensive ai use could prevent those attacks
but advocacy for that remedy is blocked by anti-ai policy/sentiment on exactly the fora that'd need to be persuaded 🫠
Choosing candidates to vote for at random, with odds according to poll results, because that makes me "impartial according to probability theorems"
(yes this is a complete butchering of the associated concepts)
(yes this is a complete butchering of the associated concepts)
Hey. If the nyt invited you to be part of a focus group what would you do as a bit?
September 26, 2026 at 1:49 PM
Choosing candidates to vote for at random, with odds according to poll results, because that makes me "impartial according to probability theorems"
(yes this is a complete butchering of the associated concepts)
(yes this is a complete butchering of the associated concepts)
Reposted by The Flaky Wanderer
Something that changed my view is the way that Suno has actively gotten me back into musical instruments.
My hands aren't built for serious practice - but I enjoy noodling around on the keyboard. Now that I can turn noodling into an actual song, I'm doing it again!
That said, speaking seriously -
My hands aren't built for serious practice - but I enjoy noodling around on the keyboard. Now that I can turn noodling into an actual song, I'm doing it again!
That said, speaking seriously -
something has flipped for me. I'm over saying "I'm not so sure" about posts like this. I am sure: the claim that LLMs eliminate all the friction that leads to learning is just wrong. don't we still have plenty to do? more than before! so there's no shortage of friction, it's just different friction
AI Coding will Prevent Expertise | Lars Faye
The need for ongoing friction in long-term skill formation.
larsfaye.com
September 26, 2026 at 7:28 AM
Something that changed my view is the way that Suno has actively gotten me back into musical instruments.
My hands aren't built for serious practice - but I enjoy noodling around on the keyboard. Now that I can turn noodling into an actual song, I'm doing it again!
That said, speaking seriously -
My hands aren't built for serious practice - but I enjoy noodling around on the keyboard. Now that I can turn noodling into an actual song, I'm doing it again!
That said, speaking seriously -
Reposted by The Flaky Wanderer
the myth of token subsidies is basically dead, right? open source inference providers are so cheap it proves the true cost of a token is very low, and the high cost of claude tokens imo indicates limited supply rather than propping up demand. oai is able to cut costs because they have more computers
September 26, 2026 at 7:39 AM
the myth of token subsidies is basically dead, right? open source inference providers are so cheap it proves the true cost of a token is very low, and the high cost of claude tokens imo indicates limited supply rather than propping up demand. oai is able to cut costs because they have more computers
Reposted by The Flaky Wanderer
This comic has been a stumbling block to many of yinz. It's a thought terminating cliché. You are not ready for it, you are not there yet.
September 26, 2026 at 10:02 AM
This comic has been a stumbling block to many of yinz. It's a thought terminating cliché. You are not ready for it, you are not there yet.
Reposted by The Flaky Wanderer
Like, the case for banning export of crypto algos was about ten orders of magnitude more strong than "but wot if word golem get too smart, wot if word golem help peepoo build bembs" and it got first-amended into oblivion regardless.
Banning open weights is an evil cause and a hopeless one
Banning open weights is an evil cause and a hopeless one
I love that the most useless spent hazbin dem of them all with a charisma of a boiled potato who someone left out in a closed lightless kettle for a week, is trying to go after open weight models based on cheap yuddite hysteria.
Like seriously, amazingly bad instincts
Like seriously, amazingly bad instincts
September 25, 2026 at 11:11 PM
Like, the case for banning export of crypto algos was about ten orders of magnitude more strong than "but wot if word golem get too smart, wot if word golem help peepoo build bembs" and it got first-amended into oblivion regardless.
Banning open weights is an evil cause and a hopeless one
Banning open weights is an evil cause and a hopeless one
Reposted by The Flaky Wanderer
copyright strikes again! 🙃
September 26, 2026 at 12:32 AM
copyright strikes again! 🙃
Reposted by The Flaky Wanderer
Copyright trolls are not your friend
So this story won't receive much attention, but... is kind of a big deal? It sure looks like the plaintiffs lawyers in the big copyright case against OpenAI/Microsoft did some seriously sketchy shit to manufacture evidence and sneak it into the case....
OpenAI Accuses Plaintiffs’ Lawyers Of Paying For, Hiding, And Then Laundering Sketchy Key Evidence In AI Copyright Case
A ton of attention...
https://www.techdirt.com/2026/09/25/openai-accuses-plaintiffs-lawyers-of-paying-for-hiding-and-then-laundering-sketchy-key-evidence-in-ai-copyright-case/
A ton of attention...
https://www.techdirt.com/2026/09/25/openai-accuses-plaintiffs-lawyers-of-paying-for-hiding-and-then-laundering-sketchy-key-evidence-in-ai-copyright-case/
September 26, 2026 at 1:07 AM
Copyright trolls are not your friend
Reposted by The Flaky Wanderer
so, are there any "philosophy of ai" papers (or papers about fundamental limits of ai/llms in general) that aren't... pseudoscience?
I think all the ones I've read went something like this:
I think all the ones I've read went something like this:
September 21, 2026 at 1:10 AM
so, are there any "philosophy of ai" papers (or papers about fundamental limits of ai/llms in general) that aren't... pseudoscience?
I think all the ones I've read went something like this:
I think all the ones I've read went something like this:
Reposted by The Flaky Wanderer
yeah I really think this is still underrated.
beats me if this is true for other neurodivergencies but for ADHD, at least the kind I’m working with, this stuff is magical
beats me if this is true for other neurodivergencies but for ADHD, at least the kind I’m working with, this stuff is magical
This is an unpopular take but as a neurodivergent self-employed person, LLM's have been an unbelievably helpful tool for the things I have struggled with my whole adult life: managing my finances, filling out paperwork, and staying on top of my email. I'm functional in a way I never dreamed of.
September 26, 2026 at 12:57 AM
yeah I really think this is still underrated.
beats me if this is true for other neurodivergencies but for ADHD, at least the kind I’m working with, this stuff is magical
beats me if this is true for other neurodivergencies but for ADHD, at least the kind I’m working with, this stuff is magical
Reposted by The Flaky Wanderer
the AI tools we have today, and therefor the fat stacks of cash the AI companies rake in, only exist because of open ML research. but now they want to pull up the ladder behind them, fence off the frontier of the field; they disparage the very idea of full public access to the latest advancements
humanity would be massively uplifted if every frontier llm was open weights. the labs do not do this because when their execs talk about wanting to uplift humanity they are ascribing themselves motives quite different from the ones that actually drive their choices
September 25, 2026 at 11:56 PM
the AI tools we have today, and therefor the fat stacks of cash the AI companies rake in, only exist because of open ML research. but now they want to pull up the ladder behind them, fence off the frontier of the field; they disparage the very idea of full public access to the latest advancements
Reposted by The Flaky Wanderer
almost like anti-ai backlash is mainly bad faith conspiracy theories and personal agendas. like, yeah, sure, a lot of executives (at ALL corporations) suck, but this shit is ridiculous.
OpenAI Accuses Plaintiffs’ Lawyers Of Paying For, Hiding, And Then Laundering Sketchy Key Evidence In AI Copyright Case
A ton of attention...
https://www.techdirt.com/2026/09/25/openai-accuses-plaintiffs-lawyers-of-paying-for-hiding-and-then-laundering-sketchy-key-evidence-in-ai-copyright-case/
A ton of attention...
https://www.techdirt.com/2026/09/25/openai-accuses-plaintiffs-lawyers-of-paying-for-hiding-and-then-laundering-sketchy-key-evidence-in-ai-copyright-case/
OpenAI Accuses Plaintiffs’ Lawyers Of Paying For, Hiding, And Then Laundering Sketchy Key Evidence In AI Copyright Case
A ton of attention was paid recently to some offhand statements from OpenAI and Microsoft employees that surfaced in filings in the NY Times' ongoing case against OpenAI, which has been consolidated into a much larger class action lawsuit. As I argued earlier, that struck me as something of a nothingburger of a story, because...
www.techdirt.com
September 25, 2026 at 11:23 PM
almost like anti-ai backlash is mainly bad faith conspiracy theories and personal agendas. like, yeah, sure, a lot of executives (at ALL corporations) suck, but this shit is ridiculous.