Sigh.
-
I absolutely agree.
At best, what current LLMs are is evidence that linguistic processing follows statistically modelable rules.
@cstross @future_upbeat
Mostly but not entirely. -
But I'm REALLY HAPPY right now because this kinda-sorta validates the key premise of the SF novel I just handed in last month (which involves serial reincarnation via destructive brain-slicing-and-imaging then imprinting onto an immature cortex, and then explores its disastrous societal failure modes).
... And it also hints that artificial consciousness might, eventually, be possible, if only via the hard path of doing it the same way we do it, only in simulation in silico.
/6 (ends)
@cstross Wait- so... I should get my brain frozen until they perfect the slicing and uploading to silicon to live eternally
-
@cstross
Also since cryogenic freezing a brain destroys the structure of an already dead brain (basically deteriotated), the folk paying for that are being scammed.I agree it's nice info for SF world building.
Presumably they'd have to replace the blood of a living mouse with a special fluid to preserve the structure?
@raymaccarthy Yes on the blood-replacement, which implies—awkwardly, for the human uploading fans—that doing this to a human would lay the experimenters open to murder charges.
-
@CGM Already noted!
-
@cstross Wait- so... I should get my brain frozen until they perfect the slicing and uploading to silicon to live eternally
@Antiqueight Naah, the ice crystals forming in your synapses would mush them into un-digitizable soup.
-
... The next step on from Drosophila, the mouse brain, is 560 times larger—never mind a vastly more complex human brain. And to get the murine connectome we'll have to chop up *a lot* of brains: a human upload won't pass any kind of medical ethics review at this point!
But near-term, it's expected to yield "fundamentally new architectural principles for AI systems that are more sample-efficient, more robust, and more capable of behavioral generalization than current approaches"
/5
@cstross > fundamentally new architectural principles for AI systems that are more sample-efficient, more robust, and more capable of behavioral generalization than current approaches
I expect Sam Altman to state that this is fake news and nobody can do things better than they do.
I also hope it'll cause panic-selling shares in all "AI" companies - but that is unlikely, it would require the average investor to have a larger brain than a Drosophila. -
Sigh.
So it turns out we've mapped the neural connectome of Drosophila *and simulated it in silico*.
Pop-sci explainer here:
Whole Brain Emulation Achieved: Scientists Run a Fruit Fly Brain in Simulation | RathBiotaClan
Scientists ran a real fruit fly brain in simulation using the FlyWire connectome, achieving the first working whole brain emulation.
RathBiotaClan (www.rathbiotaclan.com)
Key quote: "The step from a complete connectome to a working computational brain model is not trivial." And there's an even more important finding in this screenshot (alt text via OCR):
"The wiring is the computation".
/1
@cstross I didn't think that was actually in doubt, was it?
-
But I'm REALLY HAPPY right now because this kinda-sorta validates the key premise of the SF novel I just handed in last month (which involves serial reincarnation via destructive brain-slicing-and-imaging then imprinting onto an immature cortex, and then explores its disastrous societal failure modes).
... And it also hints that artificial consciousness might, eventually, be possible, if only via the hard path of doing it the same way we do it, only in simulation in silico.
/6 (ends)
@cstross Also one step closed to proving that we're likely living in a simulation.
-
But I'm REALLY HAPPY right now because this kinda-sorta validates the key premise of the SF novel I just handed in last month (which involves serial reincarnation via destructive brain-slicing-and-imaging then imprinting onto an immature cortex, and then explores its disastrous societal failure modes).
... And it also hints that artificial consciousness might, eventually, be possible, if only via the hard path of doing it the same way we do it, only in simulation in silico.
/6 (ends)
@cstross
Certainly a more promising avenue towards AGI than stochastic parrots.But then again, what they're doing here is copying a fly brain into a silicon black box and seeing what it does. The research has nothing to do with improving upon fly intelligence and immanentising the Fly Nerd Rapture.
-
Sigh.
So it turns out we've mapped the neural connectome of Drosophila *and simulated it in silico*.
Pop-sci explainer here:
Whole Brain Emulation Achieved: Scientists Run a Fruit Fly Brain in Simulation | RathBiotaClan
Scientists ran a real fruit fly brain in simulation using the FlyWire connectome, achieving the first working whole brain emulation.
RathBiotaClan (www.rathbiotaclan.com)
Key quote: "The step from a complete connectome to a working computational brain model is not trivial." And there's an even more important finding in this screenshot (alt text via OCR):
"The wiring is the computation".
/1
I heard when they first got the fly simulation up and running it introduced itself as Elon Musk and said that it was going to set up a colony on Mars.
-
@cstross I didn't think that was actually in doubt, was it?
-
But I'm REALLY HAPPY right now because this kinda-sorta validates the key premise of the SF novel I just handed in last month (which involves serial reincarnation via destructive brain-slicing-and-imaging then imprinting onto an immature cortex, and then explores its disastrous societal failure modes).
... And it also hints that artificial consciousness might, eventually, be possible, if only via the hard path of doing it the same way we do it, only in simulation in silico.
/6 (ends)
@cstross Let us know when/where the book is published. It sounds fascinating.
-
I heard when they first got the fly simulation up and running it introduced itself as Elon Musk and said that it was going to set up a colony on Mars.
@jmcrookston Yes, but the simulation model <did> set up the colony on Mars. @cstross
-
... The next step on from Drosophila, the mouse brain, is 560 times larger—never mind a vastly more complex human brain. And to get the murine connectome we'll have to chop up *a lot* of brains: a human upload won't pass any kind of medical ethics review at this point!
But near-term, it's expected to yield "fundamentally new architectural principles for AI systems that are more sample-efficient, more robust, and more capable of behavioral generalization than current approaches"
/5
Lobsters... 🦞
-
@cstross
Certainly a more promising avenue towards AGI than stochastic parrots.But then again, what they're doing here is copying a fly brain into a silicon black box and seeing what it does. The research has nothing to do with improving upon fly intelligence and immanentising the Fly Nerd Rapture.
@mrundkvist @cstross please do not give the flybros any ideas…
-
But I'm REALLY HAPPY right now because this kinda-sorta validates the key premise of the SF novel I just handed in last month (which involves serial reincarnation via destructive brain-slicing-and-imaging then imprinting onto an immature cortex, and then explores its disastrous societal failure modes).
... And it also hints that artificial consciousness might, eventually, be possible, if only via the hard path of doing it the same way we do it, only in simulation in silico.
/6 (ends)
@cstross It reminds me of something I read about 30 years ago by some Linux journalist about modelling part of the digestive ganglion of a lobster.
I wonder what happened to that guy? Not seem him in the Linux world in years...
-
Sigh.
So it turns out we've mapped the neural connectome of Drosophila *and simulated it in silico*.
Pop-sci explainer here:
Whole Brain Emulation Achieved: Scientists Run a Fruit Fly Brain in Simulation | RathBiotaClan
Scientists ran a real fruit fly brain in simulation using the FlyWire connectome, achieving the first working whole brain emulation.
RathBiotaClan (www.rathbiotaclan.com)
Key quote: "The step from a complete connectome to a working computational brain model is not trivial." And there's an even more important finding in this screenshot (alt text via OCR):
"The wiring is the computation".
/1
@cstross The popsci writeup stopped me in my tracks at the second paragraph
"The first successful polymerase chain reaction was run in a car on a California highway." Certainly not! PCR was thought out during a car drive, a *very* different thing!
https://en.wikipedia.org/wiki/Polymerase_chain_reaction#cite_ref-Mullis_97-0
-
@cstross and while hypothetically one could potentially prolong this with intensive, continuous mental health treatment? It won't succeed, because it literally can't succeed. Unavoidably at some point you have to address the facts of the matter. Which is that they are effectively just instructions on processors, and the possibility of returning to their prior body - or any truly autonomous capability - just doesn't exist.
And now you have a system with severe psychosis and homicidal urges. -
@cstross and while hypothetically one could potentially prolong this with intensive, continuous mental health treatment? It won't succeed, because it literally can't succeed. Unavoidably at some point you have to address the facts of the matter. Which is that they are effectively just instructions on processors, and the possibility of returning to their prior body - or any truly autonomous capability - just doesn't exist.
And now you have a system with severe psychosis and homicidal urges.@rootwyrm I predict that you're going to love my next novel (the one my agent's looking at right now—a few months late due to writing with cataracts).
-
@cstross I did, in fact. Said fly exists wholly within a simulated universe with limited sensor perception and no interaction with the 'real' world.
If you want useful or workable output from any sort of machine intelligence, interaction with the 'real' world is inevitable. Doubly so higher orders which may quickly key in to manipulated 'events.' Nevermind the computational requirements.
And once you cross that line, welp. Now you've got Marvin + Skynet.
