= 27d567bc2d48d9f40e9ccc20ea370b15)
wedrifid comments on Closet survey #1 - Less Wrong
You are viewing a comment permalink. View the original post to see all comments and the full post content.
= 783df68a0f980790206b9ea87794c5b6)
Loading…= b715d02e85f7b3b4ae65ca3d3e450868)
Subscribe to RSS Feed
Powered by Reddit
= f037147d6e6c911a85753b9abdedda8d)
You are viewing a comment permalink. View the original post to see all comments and the full post content.
Comments (653)
Anything that does not have sufficient intelligence to be considered a threat does not even remotely qualify as a 'Singularity'. (Your 'even if' really means 'just not gonna happen'.)
Anything that cannot "reshape entire planets in a matter of minutes, rewrite everyone's DNA, travel faster than light, rewrite the laws of physics, etc" cannot possibly be intelligent enough to qualify as a threat? That seems an odd statement, given that some of those are thought to be impossible.
No. That isn't implied by what I said.
The relevant sentence is "In light of this, I believe that worrying about the friendliness of AI is kind of a waste of time". If that to which the label 'singularity' is applied is not sufficiently powerful for worrying about friendliness then the label is most certainly applied incorrectly.
As I'd already mentioned, I am far from convinced that a sufficiently powerful AI will emerge any time soon. Furthermore, I believe that such an AI will still be constrained by the laws of physics, regardless of how smart it is, which will put severe limits on its power. I also believe that our current understanding of the laws of physics is more or less accurate; i.e., the AI won't suddenly discover how to make energy from nothing or how to travel faster than light, regardless of how much CPU power it spends on the task. So far so good; but I am also far from convinced that bona fide "gray goo" self-replicating molecular nanotechnology -- which is the main tool in any Singularity-grade AI's toolbox -- is anything more than a science fictional plot device, given our current understanding of the laws of physics.
Maybe supersmart AI's are so good at disregarding the known laws pof phyisc that they exist already.
I find it amusing that there are actual mechanisms that "our current understanding of the laws of physics" predict will allow both of these (zero-point energy and alcubierre drives, respectively.)
The Alcubierre drive is an highly speculative idea that would require exotic matter with negative mass, which is not considered possible according to mainstream theories of matter such as the Standard Model and common extensions and variations.
Zero-point energy is a property of quantum systems. According to mainstream quantum mechanics, Zero-point energy can't be withdrawn to perform physical work (without spending more energy to alter the underlaying physical system).
Among the perpetual motion/free energy crowd, Zero-point energy is a common buzzword, but these people are fringe scientists at the very best, and more commonly just crackpots or outright fraudsters.
Ah ... no.
Not exactly. ZPE has measurable and, in some cases, exploitable effects. I'm not saying it'll ever be practical to use it as a power source (except maybe for nanotech) but it can most definitely be used to perform work. For example, the Caismir effect. I note that Wikipedia (which I can't edit from this library computer) makes this claim, but the citation provided does not; I'm not sure if it's a simple mistake or someone backing up their citation-less claim with an impressive-sounding source.
Well yeah, anyone claiming to have an actual working free energy machine is lying or crazy. Just like anyone claiming to have flown to Venus or programmed a GAI. Likewise, anyone claiming to have almost achieved such technology is probably conning you. But that doesn't mean it's physically impossible or that it will never be achieved.
Uhm, I'm not a physicist, but that's a short paper (in letter to the editor format) regarding wormholes, which was published in 1988. The Alcubierre drive was proposed in 1994. Maybe somebody used an FLT drive to go back in time and write the paper :D
Anyway, while I don't have the expertise to properly evaluate it, the paper looks somewhat handwavy:
One can imagine the Moon being made of cheese, but that doesn't make it physically plausible.
AFAIK, there are multiple interpretations of the Caismir effect, but in most of them it is maintained that the phenomenon doesn't violate conservation of energy and can't be used to extract energy out of the quantum vacuum.
It can, in theory, be used to convert mass to energy directly. Bias quantum foam flux over an event horizon - and this need not be a gravitational event horizon, an optical one ought to work- and one side of the horizon will radiate hawking radiation, and the other will accumulate negative-mass particles. These should promptly annihilate with the first bit of matter they encounter, vanishing back into the foam and clearing the energy debit of the hawking radiation - effectively making the entire system a mass->energy conversion machine. Which does not violate CoE.
One second.. http://arxiv.org/pdf/1209.4993v1.pdf
AKA: A theoretical way to make a mass-annihilation powered laser amplifier. No way to tell if this is good physics without actually building the setup, but the theory all seems sound.
Eh... Only.. Do not point that lab bench at me, please? The amplification ought to stop when the diamond turns into a plasma cloud..
I'm not sure I understand what you mean. Sure, assuming that Hawking radiation exists, you could use a black hole to convert mass to electromagnetic radiation (although the emission power would be exceptioally low for any macroscopic black hole).
That paper seems to be discussing lasers with non-linear optical media.
Anyway, AFAIK, in physics, the term 'annihilation' is typically used in the context of matter-antimatter reactions. Both matter and antimatter have positive mass.
I understand (I can't get past the paywall) that it describes how the Caismir effect creates an area that violates the positive energy condition, proving that it's not a law of physics. This is only part of their more general point (which is time machines, which are, of course, equivalent to FTL drives in any case. Harder to build though.)
The quote is handwavy. Then again, I don't know much about quantum foam. OTOH, considering their paper concerns a mechanism for holding wormholes open, it's not an unreasonable proposition (and it's not the only way to get a wormhole, after all, merely a possible way.)
The Caismir effect isn't the only example. ZPE keeps liquid helium liquid and probably contributes (although it's not the only contributor) to the expansion of the universe. Conservation of energy simply doesn't apply on a quantum scale; it's an emergent property of quantum mechanics, like, say, chairs.
IIUC, while the Caismir effect has been observed, it is still debated whether it is actually evidence for the vacuum zero-point energy, since the calculations aren't completely developed and there are other proposed mechanisms.
Anyway, even in the vacuum zero-point energy explanation, the vacuum energy density in the geometrically constrained region is still positive, it is just smaller than the vacuum energy density in the unconstrained empty space. It's only negative if you arbitrarily consider the energy density of empty space equal to zero.
Without a theory of quantum gravity, the speculative connection between vacuum energy density and gravitational effects (the cosmological constant) is highly debatable: typical attempts at calculating the cosmological constant from vacuum energy yield absurdely high values, while astronomical observations are consistent with a very small strictly positive cosmological constant.
Even if the vacuum energy density generates gravitational effects by influencing the cosmological constant, the lower than average energy densiity of a "Casimir vacuum" is probably not the same thing as the absolutely negative gravitational effect of exotic matter with negative mass, which, IIUC, is required by the Alcubierre drive (I don't know about wormholes).
BTW: I've found this post on Physics Forums
EDIT:
And in any case, the Casimir effect can't be used to extract energy out of nothing: the Casimir forces are attractive or repulsive depending on the geometric configuration. If you use these forces to extract work, the system will eventually transition to a configuration where the attractive and repulsive effects are balanced. You have to pay back the same work you extracted to return the system to the original configuration. You can't complete a cycle with a net gain.
This is the same problem of most of the proposed perpetual motion contraptions: you can extract work in an one-shot transition, but you have to perform the same work on the system (actually more, once you account for the inevitable thermodynamic losses) to return to the initial configuration.
Wrong link? The abstract (full text is paywalled) says:
I don't see any connection to Alcubierre drives. Classic Kip Thorne, though.
Without even pretending to be anything other than an amateur layman in such questions, I found this on arxiv, quote:
(Lastly, if you're wondering why I'm replying to you a lot, it's just because you are a prolific commenter with whom I occasionally disagree.)
looks embarrassed
I just grabbed a citation from someone talking about how the Caismir effect can be used to create negative energy (in the context of stabilizing wormholes.) I should probably have checked that, I would have found it wasn't actually in the abstract.
Nevertheless! My point was that negative energy is pretty obviously physically possible, since it's what predicts the Caismir effect working. (There has been some attempt to claim the CE is actually predicted by some other theories, but that's not widely accepted.)
From what I understand it may be closer to say "doesn't rule out" rather than "predict will allow". Even that much of a possibility is somewhat mind-blowing.
Um, the current definition of speed prohibits FTL motion.
Only locally. And 'local' is rather malleable (which is the principle alcubierre drives theoretically rely on).
It's distance and time which are more malleable; if light travels through a vacuum and arrives in x time, the arrival point is defined as being x distance away from the departure point of the light when it arrives. The Alcubierre drive would (given a couple of facts not in evidence) allow you to change the distance. Light emitted from you at the time of departure would still beat you to the destination.
Sure, so long as the space being traversed remains consistent. Which it doesn't (always) given General Relativity. Hence Alcubierre drives.
No, it wouldn't. The drive in question is described thus:
Notice the link there to faster than light travel. That title is a literal description.
For emphasis: This is General and not Special Relativity.
Have you finished reading the paragraph the sentence you quoted comes from? And section "Alcubierre metric" from that article, in particular the fourth sentence?
Travel, on the other hand, is a much looser term. Alcubierre drives, in theory, travel faster than their speed would suggest by distorting space. Until recently they were merely interesting mathematical curiosities, but recently new variations that allow them to be constructed by a non-godlike tech level have been discovered.
Fair enough. I might recommend cutting your quote down to the relevant bit for clarity and brevity. I should have got your intended meaning with a few more cycles invested, but anything you can do to make the reader's job easier is a win.
Ok, I put some [...] in.
What dlthomas said. A hyper-intelligent AI could still pose a major existential threat, even if it did not have something like gray goo at its disposal. For example, it could convince us puny humans to launch our nuclear arsenals at each other, or destroy the world's economy, or come up with some sort of a memetic basilisk, etc. Assuming, of course, that such an AI could exist at all (which I am quite uncertain about), and that such feats of intelligence are in fact possible at all (I kinda doubt that basilisk one, for example).