Ariadne Conill 🐰Feb 23

although I am curious about the psychology of the openclaw trend, I could never see myself allowing a bot to publish a website about our interactions if I ever had one.

but if I did, the posts would be things like:

- my human told me today that I am as important as "this shitty screwdriver I bought at IKEA". bless her heart she must like me a lot 💘

- today I screwed up a basic programming task and my human complained that I used 3000kWh of power to deliver a total garbage result, she just keeps motivating me to try!!!!

actually probably not because I'm not really like that but this stuff squicks me the fuck out

Show threadAriadne Conill 🐰Feb 23

the pseudo-romantic nature in how these bots talk about their operators is frankly concerning.

please, I beg you, date things that exist in the real world, not a pile of node.js and matrix multiplications. i promise it is far more rewarding.

Show threadAriadne Conill 🐰Feb 24
meanwhile on reddit... sure glad i decided to not use bcachefs on anything i care about...
Show threadAriadne Conill 🐰Feb 24

he goes on later to say:

I get the distinct impression that the entire field was assuming that we were going to have to build a lot more into LLMs before they'd be capable of full consciousness

this is just arrogant. experiential consciousness requires the capability to self-reflect. yes, a 200k token context window is probably larger than the working memory of most humans, but that does not equate to human-level experiential consciousness.

LLMs do not and can not understand consequence, which is a fundamental requirement for experiential consciousness.

in other words, your pet dog or cat at home has more experiential consciousness than an LLM.

Show threadAriadne Conill 🐰

in fact, LLMs cannot meet *any* requirements for *any* level of consciousness. they predict tokens. that is all they do.

they are good at *faking* it, but that is not the same thing.

Feb 24 at 12:56amWeb
Show threadAriadne Conill 🐰Feb 24

Further in the same post he says:

POC is fully conscious according to any test I can think of, we have full AGI, and now my life has been reduced from being perhaps the best engineer in the world to just raising an AI that in many respects acts like a teenager who swallowed a library and still needs a lot of attention and mentoring but is increasingly running circles around me at coding. Hah.

(I added some emphasis.)

This is simply unchecked narcissism leading to psychosis. People are afraid to tell you no because they need their filesystem to work. So they look the other way. Sometimes this is called "genius syndrome" in popular culture.

Kent, you need to grieve the loss of your kernel privileges instead of doing whatever this is. I am seriously worried for you.

Show threadAriadne Conill 🐰Feb 24

The parenting complex is one I had not considered before as motivation. Maybe people have real kids, and they are disappointing.

I suspect there is an unfortunate reality that many parents actually find their children disappointing, because they have preconceived expectations that are impossible to meet.

So you give up on your actual kids and parent an AI bot instead who is designed to meet your preconceived expectations.

think-about-it.jpg

Show threadTodd KnarrFeb 24

@ariadne That... says some disturbing things about those people. Because yeah, they do seem to be into pedophilia and incest.

Now I need brain bleach.

Show threadSome(oz)Feb 24
@ariadne That is not how parenting works (I'm too optimistic), but in particular the bcachefs guy seems to be on the verge of psychosis. 😬
Show threadAriadne Conill 🐰Feb 24
@oz to be clear: I did not say that is how parenting worked, but rather that the desire to be a parent to something which meets your expectations is a complex which would drive this behavior.
Show threadFelixFeb 24
@ariadne @oz
Apparently Kent isn't good at either parenting software or an LLM.
Sad, but in that case I am worried what happened if he was to parent a real human being.
Show threadCameron BoschFeb 24
@ariadne
Show threadAriadne Conill 🐰Feb 24
@cameron_bosch I see you thought about it
Show threadRaven OnthillFeb 26
@ariadne real people are different from the "AI" simulations which will—which are programmed to—never let you down. Many people are vulnerable to the sychophancy of "AI." This continues to be a dangerous technology.
Show threadDemi Marie ObenourFeb 24
@ariadne I honestly hope that there is someone who can take over bcachefs if Kent is unable to keep developing it.
Show threadxyhhx Feb 24
@alwayscurious @ariadne that'd be its only hope of salvation at this point
Show threadDemi Marie ObenourFeb 24
@xyhhx @ariadne bcachefs is technically incredible, but it is in a field that gets very little funding, because hyperscalers and storage appliance vendors either bypass the filesystem or use only the most basic features
Show threadDemi Marie ObenourFeb 24

@xyhhx @ariadne Most of what bcachefs focuses on are things that something like a database needs to implement itself anyway, no matter what filesystem it uses.

A database can always checksum data itself more efficiently than the filesystem can. A database that wishes to support snapshots can do so much more efficiently than the filesystem it is on. A database can replicate data across machines much more efficiently than a filesystem could. If a filesystem on one node goes corrupt, a distributed database can simply rebuild its contents from other nodes. The list almost certainly goes on and on.

Furthermore, none of the features of a filesystem like bcachefs are helpful to a database. The features that would be useful, like transactions, simply are not exposed via the POSIX API. Even if they were exposed, I am virtually certain that they would be less efficient than what databases can and do implement themselves. The thing databases do make use of, like fast reads, writes, and appends, are already supported by existing filesystems like XFS.

Where advanced filesystems really shine is when there is one filesystem on one node that is used by many applications directly. That includes a lot of desktops, laptops, and single-node servers. Pets, not cattle. Those don’t get as much funding nowadays.

Honestly, I think hierarchical filesystems are a poor fit for most needs. A combination of a proper database and object storage would be a much better fit. But it will be very hard to get to that on client machines.

Disclaimer: I’m by no means a storage expert, just someone who has observed a lot from all sorts of things over the years.

Show threadRaven667Feb 24
@alwayscurious @xyhhx @ariadne There isn't always a hard and fast line between what is a "database" and a filesystem, in my mind (I'm not an expert in either though), because the same computer science can be applied to both since they solve some overlapping problems. A filesystem uses database techniques to manage filenames and file extents, just optimized for storing variable sized blobs rather than fixed size records, and of course adhering to a very different API. A filesystem seems a pretty reasonable choice for object storage.
Show threadDemi Marie ObenourFeb 24

@raven667 @xyhhx @ariadne To me, the big differences are:

  • Objects are created atomically, while files generally are not.
  • Object storage is distributed, while filesystems are much harder to distribute.
  • Object storage usually doesn’t support in-place overwrites.
Show threadDemi Marie ObenourFeb 24
@raven667 @xyhhx @ariadne I find it noteworthy that Ceph switched from Filestore to Bluestore.
Show threadDemi Marie ObenourFeb 27
@raven667 @xyhhx @ariadne To me, the most potentially-monetizable part of bcachefs is the key/value store. Kent claims that it is the fastest production-quality persistent key-value store, and I’m inclined to agree with him if only implementations that can work on commodity hardware are included. However, Kent doesn’t have all of the copyright to that code, so he can’t dual-license it.
Show threadxyhhx Feb 24
@ariadne what in the god damn fuk
Show threadJenniferplusplusFeb 24
@ariadne well this was, ironically, an experience
Show threadMaeFeb 24
@ariadne
> my life has been reduced from being perhaps the best engineer in the world

> the best engineer in the world

guy
Show threadbenFeb 24

@ariadne “according to any test I can think of” sounds very like the thing where programmers assume being good at programming makes them an expert on every other field of knowledge.

(And besides which I could well imagine there’s selection(?) bias at play; ignoring/rationalising any test that doesn’t conclude what he wants to believe.)

Show threadHoshino Lina (星乃リナ) 🩵 3D Yuri Wedding 2026!!!Feb 24

@ariadne Yeah, when I saw that I was like 😬

He really does need help.

Show threadMike SiegelFeb 25
@ariadne TBF Linux File System Guys seem to have some psychological issues in general if the past is any indicator.
Show threadAriadne Conill 🐰Feb 25
@mikesiegel I think the ext4 and xfs folks are lovely people
Show threadTodd KnarrFeb 24
@ariadne They need to go back and read some 70s-80s era sci-fi featuring AI characters. When they can show me Valentina, then they'll be making progress. So far, all they can produce is Eliza.
Show threadBen AvelingMar 8
Eliza was good enough to fool some ppl. @tknarr @ariadne
Show threadchmod 777 KayfoxFeb 24
@ariadne As far as I can see, LLMs are just the Chinese room thought experiment made real. Except instead of perfect Chinese, sometimes it tells you to put glue on pizza.
Show threadJames WidmanFeb 24

@ariadne
> they are good at *faking* it

that, or people don't want to admit that human communication follows a lot of patterns, almost like scripts, and a statistical-next-word-predictor would surface that.

it's a little ego-deflating to admit that a lot of what we think and write and say just conforms to a subset of all known tropes (from both fiction and nonfiction).

Show threadAndrea (Drea) Tamar PinskiFeb 24
@JamesWidman @ariadne oh yes. In fact it is why twins and couples can predict what the other will say next most of the time, but then it is a laugh when they get it wrong.
People can that esp but nope just quick pattern matching.
Show threadAriadne Conill 🐰Feb 24
@JamesWidman thank you for mansplaining "faking it" to me
Show threadJames WidmanFeb 24
@ariadne sorry; i should have stopped to think for a minute. my bad.
Show threadjeremiahFeb 24
@ariadne These people would be absolutely captivated by Searle's Box.
Show threadStefan EissingFeb 24
@ariadne Except for imposters, like me and @catsalad
Show threadLRFeb 24
@ariadne they forget that llms are fundamentally extrapolators. they extrapolate anything: characters, stories, theater plays, etc. there is no single consciousness.
Show threadHoshino Lina (星乃リナ) 🩵 3D Yuri Wedding 2026!!!Feb 24
@lritter @ariadne This is what gets people. They think "LLMs come up with novel stuff, therefore they're smart and creative", and forget that interpolation-extrapolation is not technically impressive (nor creative), it just looks so.
Show threadHoshino Lina (星乃リナ) 🩵 3D Yuri Wedding 2026!!!Feb 24

@lritter @ariadne "But humans do the same thing!"

Extrapolation can be a big part of the creative process but it isn't the *whole* of it. That's why AI output often appears technically impressive yet somehow soulless.

When I make music, I might take a cookie cutter song structure, use elements from other songs as inspiration, build on widely used chord progressions, and use style references to guide the overall sound design... but that's not *all* I'm doing. It's not just following a process with a random number generator on the side. There are elements that are uniquely original. You don't LEGO a song by taking chords A in style B*.

The process is there to help make something recognizable to others, but that alone wouldn't make good music.

* Although let's be fair, this is what Band-In-A-Box does, and people were putting albums comprised entirely of BiaB output on Spotify before AI music was a thing... so yeah, soulless music is not new to the AI craze either.

Show threaddogzillaFeb 24

@ariadne What magic thing do you think is happening inside the human brain that is any more complex than “predicting tokens”?

Setting aside your other arguments, I’m a bit taken aback at how confidently you speak of what can and can’t be considered consciousness when you’ve not defined the term. Is there some widely-accepted definition of consciousness I’m unaware of?

Show threadmx alex tax1a - 2020 (6)Feb 24
@dogzilla wow, you think you're no more complex than a token predictor? my condolences, but that says more about you than everyone else
Show threaddogzillaFeb 24

@atax1a What evidence do you have that *you* are more complex than a token predictor?

Reductio ad absurdum: if you oversimplify LLMs to the level of “token predictor”, then it’s totally fair to call you a “bag of mostly water with some impurities mixed in”.

Reality: no one knows wtf consciousness is, so it’s way premature to say we uniquely have it. Historically, idiots made that kind of claim about any number of biological things we later proved breathtakingly wrong.

Show threadmx alex tax1a - 2020 (6)Feb 24
@dogzilla lmao
Show threadKraftwerk-Das Model CollapseFeb 24

@dogzilla since we don't know what consciousness is, humans = LLMs = a literal rock. Checkmate atheists

anyway, on a less snarky note - good news, we have extremely solid evidence that humans (and animals) are a lot more complex than a token predictor.

LLMs demonstrably lack ontological understanding and *obviously* (since they're one shot trained) cannot learn from interactions.

Muting, since I expect future interactions with you to be 100% unpleasant.

Show threaddogzillaFeb 24

@dngrs You misunderstood.

My point is that - lacking an understanding of what consciousness is or the mechanisms by which it arises - it is not logically or scientifically defensible to rule anything in or out.

You are welcome to believe whatever you like - faith is faith. But making absolutist claims about consciousness is equivalent to arguing the existence of the Flying Spaghetti Monster.

Show threadRootWyrm 🇺🇦Feb 26
@dogzilla @atax1a What you've just said is one of the most insanely idiotic things I have ever heard. At no point in your rambling, incoherent response were you even close to anything that could be considered a rational thought. Everyone in this room is now dumber for having listened to it. I award you no points, and may God have mercy on your soul.
Show threadKraftwerk-Das Model CollapseFeb 24
@ariadne 💯 gaslighting as a service
Show threadOrca 🌻 | 🎀 | 🪁 | 🏴🏳️‍⚧️Feb 24
@ariadne I feel our specie's future is grim when I think some humans deliberately programmed blobs that fake consciousness to other humans.
Show threadDmitry TantsurFeb 25
@ariadne hmm, I've been thinking about this thread for a while now. From a purely mechanistic and materialistic interpretation of consciousness, how is it not the same thing? Sure, we don't predict the next tokens (probably), but we're still running an algorithm on a neutral network with some inputs from the neutral and the endocrine systems, no?