|
Polybius91 posted:What I like about that Yud quote is that the easy part is going to be building an AI that isn't hilariously awful at image recognition. Which has been one of the major hurdles of computer science since forever. Shut up AI go FOOM.
|
#
?
Aug 7, 2015 00:36
|
|
|
|
| # ? May 8, 2024 09:01 |
|
|
Sham bam bamina! posted:Oh lordy. It was on Twitter. He was probably tweeting drunk. Hazardous occupation, trying to be funny when you're pathologically humorless. The Vosgian Beast posted:tbf, Yudkwosky is more the person they really really want to join their lunch table than a permanent occupant WELL THEY SURE hosed THAT ONE DIDN'T THEY. BobHoward posted:Just look at this gem of a Yud quote su2su2u1 dug up. Now now, it's clearly intended as Science Fiction to instil a Sense of Wonder. With HPMOR he was still writing SF/F, just more F. Think of your MIRI donation as a Patreon for this ongoing literary work. BobHoward posted:Super intelligence ain't gonna compensate for the fact that the signal you're looking for is waaaaay below the noise floor of that sensing system, especially when you're loudly proclaiming super AI will need only one data point. This is also the MIRI argument for cryonics. "But a sufficiently powerful AI ..." "No. It's hosed. You hosed it. The data is gone." "But a SUFFICIENTLY POWERFUL AI ..." The Vosgian Beast posted:You get in there as a teen who just became an atheist, and then you're too emotionally attached to its role in your life to ever really leave, even as you begin to admit more and more of Yud's writings are garbage. I started reading it at 43. Told you "child genius adult moron."
|
|
#
?
Aug 7, 2015 01:18
|
|
|
divabot posted:This is also the MIRI argument for cryonics. "But a sufficiently powerful AI ..." "No. It's hosed. You hosed it. The data is gone." "But a SUFFICIENTLY POWERFUL AI ..." My favorite part of this is even IF you had an AI that could reconstruct parts of a person based on archived Facebook posts and crap, there's going to be fundamental limits on how accurate a picture of a person you can reconstruct, because entropy always takes its pound of flesh. It is also nice to see someone trying to explain that, no, neurons do not work like logic gates.
|
|
#
?
Aug 7, 2015 01:46
|
|
|
My robot god is a mighty god.
|
|
#
?
Aug 7, 2015 01:57
|
|
|
divabot posted:This is also the MIRI argument for cryonics. "But a sufficiently powerful AI ..." "No. It's hosed. You hosed it. The data is gone." "But a SUFFICIENTLY POWERFUL AI ..." Cyronics is homeopathy for the educated idiot. Despite none of the original substance remaining, crynicists believe
|
|
#
?
Aug 7, 2015 02:00
|
|
|
Like that flying soda can in Destiny that scans random dead skeleton russians and makes player characters.
|
|
#
?
Aug 7, 2015 03:42
|
|
|
Hate Fibration posted:My favorite part of this is even IF you had an AI that could reconstruct parts of a person based on archived Facebook posts and crap, there's going to be fundamental limits on how accurate a picture of a person you can reconstruct, because entropy always takes its pound of flesh. It is also nice to see someone trying to explain that, no, neurons do not work like logic gates. Can't wait for the AI that will reconstruct me based on the dozens-to-hundreds of social media posts I made while under the impression that I was straight. 
|
|
#
?
Aug 7, 2015 04:42
|
|
|
DStecks posted:Can't wait for the AI that will reconstruct me based on the dozens-to-hundreds of social media posts I made while under the impression that I was straight. Supremely perfect AI know that posthumanist heaven-on-earth will need everyone reconstructed in perfect form: straight, white, technocratic, and libertarian. You will be remade in rationalist, social-darwinist perfection and spend the eons arguing with Yud about information theory and cryonics, even though neither of you has ever so much as taken a college course in neurology. You will see nothing wrong with that, now that your brain has been perfected by the subtle modifications needed to turn you into a bloviating shitheel droning on forever and ever after about eugenics and the glory of the market. And you won't want it any other way. You'll see nothing wrong with it because you'll love it then in Heaven forever and ever.
|
|
#
?
Aug 7, 2015 04:52
|
|
|
Jack Gladney posted:Supremely perfect AI know that posthumanist heaven-on-earth will need everyone reconstructed in perfect form: straight, white, technocratic, and libertarian. You will be remade in rationalist, social-darwinist perfection and spend the eons arguing with Yud about information theory and cryonics, even though neither of you has ever so much as taken a college course in neurology. You will see nothing wrong with that, now that your brain has been perfected by the subtle modifications needed to turn you into a bloviating shitheel droning on forever and ever after about eugenics and the glory of the market. The indie survival horror game Master Reboot is all about how much of a nightmare this concept is when you think about it more than two seconds.
|
|
#
?
Aug 7, 2015 05:28
|
|
|
Man, I think robot hookers are a lost more feasible way to find a waifu that wants to gently caress you.
|
|
#
?
Aug 7, 2015 05:29
|
|
|
There's an interesting Black Mirror episode about a service that reconstructs dead people from social media posts.
|
|
|
#
?
Aug 7, 2015 05:34
|
|
|
Jack Gladney posted:Supremely perfect AI know that posthumanist heaven-on-earth will need everyone reconstructed in perfect form: straight, white, technocratic, and libertarian. You will be remade in rationalist, social-darwinist perfection and spend the eons arguing with Yud about information theory and cryonics, even though neither of you has ever so much as taken a college course in neurology. You will see nothing wrong with that, now that your brain has been perfected by the subtle modifications needed to turn you into a bloviating shitheel droning on forever and ever after about eugenics and the glory of the market. "Aumann's agreement theorem, roughly speaking, says that two agents acting rationally (in a certain precise sense) and with common knowledge of each other's beliefs cannot agree to disagree. More specifically, if two people are genuine Bayesians, share common priors, and have common knowledge of each other's current probability assignments, then they must have equal probability assignments."
|
|
#
?
Aug 7, 2015 05:50
|
|
|
chrisoya posted:Hang on, I thought sufficiently-rational people automatically agreed, proven by Aumann's Agreement Theorem? If you're arguing then you're not smart enough, get out of Yud's heaven you fake Bayesian A classic deepity.
|
|
#
?
Aug 7, 2015 06:01
|
|
|
DStecks posted:Can't wait for the AI that will reconstruct me based on the dozens-to-hundreds of social media posts I made while under the impression that I was straight. It's funny because historians have been arguing for centuries about how much you can really get from a personal account of anything, since literally everything a person leaves behind is written with some sort of audience in mind (even if it is just yourself), which you have to take into account when you interpret a source and which means that there are no sources that objectively represent what a person really thinks. At least not to the degree that full personality reconstruction would require. But I guess since a sufficiently advanced AI is indistinguishable from God that won't be a problem.
|
|
#
?
Aug 7, 2015 07:00
|
|
|
chrisoya posted:Hang on, I thought sufficiently-rational people automatically agreed, proven by Aumann's Agreement Theorem? If you're arguing then you're not smart enough, get out of Yud's heaven you fake Bayesian Actually, that's ( a perversion of) something Socrates said.
|
|
#
?
Aug 7, 2015 09:52
|
|
|
"First, assume all participants in a conversation are frictionless spheres in a vacuum... "
|
|
#
?
Aug 7, 2015 15:25
|
|
|
Remember that you will be mind-killed if you have any sort of actual values. However it is impossible not to have values if you have any awareness of other humans, so your values by default are those that maintain the societal status quo and prevent social change. And people wonder why nerdlords are so conservative.
|
|
#
?
Aug 7, 2015 15:43
|
|
|
Hate Fibration posted:My favorite part of this is even IF you had an AI that could reconstruct parts of a person based on archived Facebook posts and crap, there's going to be fundamental limits on how accurate a picture of a person you can reconstruct, because entropy always takes its pound of flesh. It is also nice to see someone trying to explain that, no, neurons do not work like logic gates. "But wait, maybe all the information that's irrevocably lost in the freezing process is stuff that doesn't matter!" -an actual argument being made in that thread
|
|
#
?
Aug 7, 2015 16:56
|
|
|
Alien Arcana posted:"But wait, maybe all the information that's irrevocably lost in the freezing process is stuff that doesn't matter!" -an actual argument being made in that thread The arguments for cryonics always involve a heavy dose of "but you can't prove it isn't true!", reversing the burden of proof. Also, nanobots.
|
|
#
?
Aug 7, 2015 17:09
|
|
|
It's weird how they imagine that "intelligence" can be essentially an unbounded process (which is the only way you can invoke "sufficiently advanced AI" to solve every conceivable problem). We know nothing about intelligence beyond our own, so why do they think that they can say anything meaningful about it?
|
|
#
?
Aug 7, 2015 17:32
|
|
|
Jack Gladney posted:Supremely perfect AI know that posthumanist heaven-on-earth will need everyone reconstructed in perfect form: straight, white, technocratic, and libertarian. You will be remade in rationalist, social-darwinist perfection and spend the eons arguing with Yud about information theory and cryonics, even though neither of you has ever so much as taken a college course in neurology. You will see nothing wrong with that, now that your brain has been perfected by the subtle modifications needed to turn you into a bloviating shitheel droning on forever and ever after about eugenics and the glory of the market. I thought the LW thread established that in this Heaven you'll also be a pony
|
|
#
?
Aug 7, 2015 18:01
|
|
|
ol qwerty bastard posted:It's weird how they imagine that "intelligence" can be essentially an unbounded process (which is the only way you can invoke "sufficiently advanced AI" to solve every conceivable problem). We know nothing about intelligence beyond our own, so why do they think that they can say anything meaningful about it?
|
|
#
?
Aug 7, 2015 18:22
|
|
|
darthbob88 posted:I've figured it would be so, simply by analogy with computers; with sufficient processing power, you can simulate every conceivable scenario to find one that would lead to either the desired end or the current state of affairs. On the other hand, as we have discussed in this thread, "sufficient" is a weasely loving word, and you also need sufficient information to construct sufficiently accurate simulations. For a computer to brute-force Super Mario Brothers, this comment notes that if you had a computer that could perform one computation per Planck time per Planck volume in the volume of the entire universe, it would take ~6 * 10^27570 years. But "unfeasible" is probably tawdry cached-thoughts domain expertise, not superior generally-applicable rationality.
|
|
#
?
Aug 7, 2015 20:32
|
|
|
ol qwerty bastard posted:It's weird how they imagine that "intelligence" can be essentially an unbounded process (which is the only way you can invoke "sufficiently advanced AI" to solve every conceivable problem). We know nothing about intelligence beyond our own, so why do they think that they can say anything meaningful about it?
|
|
#
?
Aug 7, 2015 20:35
|
|
|
ikanreed posted:"First, assume all participants in a conversation are frictionless spheres in a vacuum... " Imagine four agents acting rationally on the edge of a cliff.
|
|
#
?
Aug 7, 2015 20:48
|
|
|
My Dark Enlightenment name is Johnny Fiveaces
|
|
#
?
Aug 7, 2015 20:51
|
|
|
Any Dark Enlightenment reactions over the Republican debate last night? I know the consensus is that politics is a mind killer, but from what I've seen these guys and their ilk are big Trump supporters.
|
|
#
?
Aug 7, 2015 21:57
|
|
|
MizPiz posted:Any Dark Enlightenment reactions over the Republican debate last night? I know the consensus is that politics is a mind killer, but from what I've seen these guys and their ilk are big Trump supporters. Just the usual masturbating over revolution.   
|
|
#
?
Aug 7, 2015 23:41
|
|
|
Looks like the three letter inquisitors will be harshing Clarkhat's mellow soon.
|
|
#
?
Aug 8, 2015 02:39
|
|
|
Ronwayne posted:Looks like the three letter inquisitors will be harshing Clarkhat's mellow soon. Man, social justice warriors are the new Torquemadas of the world. *writes a long blogpost defending the Spanish Inquistion*
|
|
#
?
Aug 8, 2015 02:41
|
|
|
God I hope so.
|
|
#
?
Aug 8, 2015 02:42
|
|
|
The Vosgian Beast posted:Man, social justice warriors are the new Torquemadas of the world. I meant the FBI but sure.
|
|
#
?
Aug 8, 2015 03:04
|
|
|
Ronwayne posted:I meant the FBI but sure. What's the details?
|
|
#
?
Aug 8, 2015 03:26
|
|
|
ianmacdo posted:What's the details? He's making a joke because Clarkhat talked about dumping ebola on major cities because he doesn't like their politics.
|
|
#
?
Aug 8, 2015 03:29
|
|
|
Rare video footage of the Dark Enlightenment meeting where Annisimov got kicked out https://youtu.be/rPv2I-mArD8
|
|
#
?
Aug 8, 2015 03:39
|
|
|
The Vosgian Beast posted:Man, social justice warriors are the new Torquemadas of the world. The Something Awful Forums > Discussion > Post Your Favorite or Request > The Dark Enlightenment: *writes a new post defending the Spanish Inquisition*
|
|
#
?
Aug 8, 2015 15:47
|
|
|
Honestly, the best argument that voting does matter, and one that might get through to these guys, is that companies are literally spending billions of dollars to influence it. Multinational companies aren't known for spending that much money on anything that won't help them.
|
|
#
?
Aug 8, 2015 15:52
|
|
|
That would require believing that something the sheeple think is important actually is important, which they aren't gonna do.
|
|
#
?
Aug 8, 2015 16:02
|
|
|
NRx is built upon incessant contrarianism in all things. Such as in believing that Trump is a genius mastermind troll as opposed to a moron.
|
|
#
?
Aug 8, 2015 17:30
|
|
|
|
| # ? May 8, 2024 09:01 |
|
|
Zemyla posted:Honestly, the best argument that voting does matter, and one that might get through to these guys, is that companies are literally spending billions of dollars to influence it. Multinational companies aren't known for spending that much money on anything that won't help them. No the best argument is "Does voting matter? Let's ask this Iraqi civilian about whether he would prefer a Bush or a Gore administration. Oh wait we can't because he's dead."
|
|
#
?
Aug 8, 2015 17:37
|
|