Open Side Menu Go to the Top
Register
"The Singularity Is Near" by Ray Kurzweil, How Close?? "The Singularity Is Near" by Ray Kurzweil, How Close??

08-07-2010 , 07:56 PM
Quote:
Originally Posted by TomCowley
How can you call the network ever-present and worldwide if I can't have it all the time wherever I go in the world? WTF? And I'm not being as ass and asking for it in Antarctica or at the bottom of the ocean. His wireless network (that he says is capable of rapidly distributing movies) doesn't even exist on the average downtown sidewalk tile, much less in less urban places.



I've already explained that the word ever refers to time, not geography. Google it. Www.dictionary.com.


As for worldwide network... omfg dude. The internet is a world-wide network. It doesn't matter if you can't connect to it from 90% of land locations. It's still a world-wide network. Has been referred to that since day 1.

Clearly the adjective world-wide is describing this network. And clearly we have networks that this adjective is fitting of.
"The Singularity Is Near" by Ray Kurzweil, How Close?? Quote
08-07-2010 , 07:57 PM
Quote:
Originally Posted by Max Raker
It depends on when it was made. Saying that about letters or whatever in the mid 80s would have been impressive. When did Kurzweil say that about music? If it was after in mid/late 90s it is barely a prediction as that was already happening.
Late '80's.

95% sure it was in tAoIM, but i lost my copy years ago so I'm going off memory.
"The Singularity Is Near" by Ray Kurzweil, How Close?? Quote
08-07-2010 , 08:27 PM
Quote:
Originally Posted by ZeeJustin
Late '80's.

95% sure it was in tAoIM, but i lost my copy years ago so I'm going off memory.
Seems like a solid prediction then
"The Singularity Is Near" by Ray Kurzweil, How Close?? Quote
08-07-2010 , 09:43 PM
Quote:
Originally Posted by Max Raker
Seems like a solid prediction then
The prediction I quoted is word-for-word from his 1999 book.
"The Singularity Is Near" by Ray Kurzweil, How Close?? Quote
08-07-2010 , 09:52 PM
Quote:
Originally Posted by ZeeJustin
Conducting experiments on the brain to learn more about it has been done since monkeys poked brains w/ sticks.

Clearly more is meant by reverse engineering than this.

It refers to analyzing the structure, which was first done by using frozen brains, and slicing them into super thin layers to be able to look at the inside of the brain on a cellular level. MRI's are not precise enough to be considered reverse engineering.
Jesus ****ing christ read what fMRI does already. You have no clue what you're talking about.



Quote:
And ya, your interpretation re digital objects is insane. He never implied that hard copies don't exist.
Lol, because "typically do not have physical objects associated with them" means that they all have physical objects associated with them. Of course. He's just too damn brilliant for me to understand that he meant the exact opposite of what he said.

You're looking at the state of the world- "people routinely buy books, software, and music (movies, not so much atm) as digital objects instead of as physical objects (or instead of on physical media)" and using horoscope / john edwards style association to think that's what he actually predicted. That's not what his prediction says. At all. Somebody else warned of this bias earlier ITT.
"The Singularity Is Near" by Ray Kurzweil, How Close?? Quote
08-07-2010 , 11:02 PM
Quote:
Originally Posted by TomCowley
Jesus ****ing christ read what fMRI does already. You have no clue what you're talking about.
fMRI's provide functional, not structural data.

There are MRI techniques that provide structural information, but they are relatively new.

fMRIs are used to examine blood flow, which (in theory) is correlated to neuronal activity. As far as I know, you can't use an fMRI to determine microstructural features.

To do that, you would likely use diffusion tensor MRI. Basically, DTMRI data shows how water flows in a sample. Structural elements like collagen exhibit anisotropic water diffusion, so you can guess structural makeup / alignment of a tissue sample.

DTMRI was not standard in the 1990s. From what I can tell from wikipedia, it wasn't prototyped for biology until ~1992-1993 (earliest pubmed paper is 94).

DTMRI usage to elucidate structure still isn't commonplace - fields like cardiology / developmental biology still haven't caught up with it, and from what I've seen, new cardiac structural information has been published as late as 2009. I'm not acquainted with its usage in neuroscience.

Pubmed shows 67 papers with "diffusion" + "tensor" + "brain," of which about 54 are published 1998-2000. Many of the prior 13 papers are concerned with what you'd expect - proof of concept, underlying physics, etc.

In contrast (lolz) the same search shows 3100 papers in the 2000s.
"The Singularity Is Near" by Ray Kurzweil, How Close?? Quote
08-07-2010 , 11:22 PM
Quote:
Originally Posted by Plancer
fMRI's provide functional, not structural data.
And when you conduct various experiments, you see in real time which specific parts of the brain are involved in which specific things. That's clearly a step towards reverse engineering, (remember, the prediction was that research would be INITIATED, not that it would be far advanced or anything). It had been initiated in 1999. Research with your technique had also already been initiated in 1999. if your citations are correct.
"The Singularity Is Near" by Ray Kurzweil, How Close?? Quote
08-07-2010 , 11:51 PM
Quote:
Originally Posted by TomCowley
...
Prediction: "Digital objects such as books, music albums, movies, and software are rapidly distributed as data files through the wireless network, and typically do not have a physical object associated with them."

I hope I don't have to diagram this sentence for you, but "them" at the end refers to "digital objects such as .. music albums..". It does NOT refer to "individual data files that are distributed through the wireless network" (which obviously don't have a physical object.. what are they going to do, mail you a physical copy too after you download it? That reading is prima facie absurd compared to mine.).

The prediction is that the ALBUM (the digital object, the master copy of the album) typically does not have a physical object associated with it. The prediction is NOT that your COPY of the album- the data file- typically does not have a physical object associated with it. The prediction is that the album itself is typically an entirely digital object that is not stored (or the weaker condition, not distributed, if you prefer), in physical form. It is obviously completely wrong.
You say in your post "them" refers to "digital objects."
Then, you say "the prediction is that the ALBUM..."
The "album" may or may not have a physical copy associated with it. The digital object never (or rarely) does.

Grammatically, this is clear - the "them" in the sentence "[is distributed] through the wireless network" - the only noun which "them" could refer is "instantiations (copies) of the album." I agree with Max Raker / ZJ that most people will adopt the pro-Kurzweil interpretation.

This interpretation of the prediction probably makes more sense when it was made than today. You call that interpretation "prima facie absurd" largely because the practice of customers purchasing completely digital objects has become mainstream.

This prediction was not obvious, and likely would have encountered a great deal of resistance / disbelief when presented ye olde simple folk.

Digital Media as a product is a very new practice, and used to be delegated to fringe geek stuff.
The iTunes store was launched in 2003. Napster was still running until 2001. The app store, and the various online stores for the gaming systems (XBox, etc) would be inconceivable to most people in the 90s.

At the time when Kurzweil wrote this prediction, the notion that there would be physical copies associated with digital copies wasn't absurd - it was actually considered by the major record labels as a way of preserving their business model while letting their users continue to enjoy mp3s. Obviously, it didn't work.

The weakness of this prediction is that it doesn't clarify whether the "distributed objects" are being distributed via legal means or piracy. Furthermore, he obviously overestimated the impact of digital books.
"The Singularity Is Near" by Ray Kurzweil, How Close?? Quote
08-07-2010 , 11:56 PM
Quote:
Originally Posted by Max Raker
Well, I can't really argue with you if you haven't done your homework. Computers are still terrible at solving games. They can't even break RSA encryption yet, much less Graph isomorphisms or known NP complete problems. After that you can start talking about EXPTIME.

And my point on the complexity being an advantage was that you could make a game complicated enough to where no human could even play. Real games involve heuristics which computers obv are not very good at yet,
Computers are so stupid! Yesterday, I bumped into one that couldn't even solve NP complete problems...let alone break RSA!

I mean, com'on, who doesn't know how to solve NP complete problems?
"The Singularity Is Near" by Ray Kurzweil, How Close?? Quote
08-08-2010 , 12:30 AM
Quote:
Originally Posted by Plancer
You say in your post "them" refers to "digital objects."
Then, you say "the prediction is that the ALBUM..."
The "album" may or may not have a physical copy associated with it. The digital object never (or rarely) does.

Grammatically, this is clear - the "them" in the sentence "[is distributed] through the wireless network" - the only noun which "them" could refer is "instantiations (copies) of the album." I agree with Max Raker / ZJ that most people will adopt the pro-Kurzweil interpretation.
Your grammar is wrong. If he wanted to refer to people's individual digital copies of albums, there were many possible ways to write that (I gave a simple one ITT). If he meant the albums themselves...well, he would have said what he did.
"The Singularity Is Near" by Ray Kurzweil, How Close?? Quote
08-08-2010 , 12:45 AM
Quote:
Originally Posted by TomCowley
You obviously don't know what fMRI is, what it is used for, and how it differs from a normal medical MRI. It's used for the express purpose of conducting experiments to learn how the brain works.
I actually disagree here.
fMRIs are good at finding out what regions of the brain are correlated with certain tasks. I don't think it really provides much information about how the brain works.

I think to qualify for reverse engineering, you need to be able to 'black-box' study it. You need to be able to provide a stimulus, measure the response, then be able to make this claim:
"If I replaced this thing with a black box with the same stimulus / response relationship, the system wouldn't notice."

fMRI doesn't provide us with this sort of information. You can do all the fMRI testing you want - you'll never be able to substitute a region of the brain (or the brain itself) with a interoperable device, nor could you design a system based on similar properties. fMRI data is a development in the spirit of the old school "ablative" neuroscience - you can find out what regions are activated for an activity, but that's all (example of old school ablative is "stroke patients who have lesions in Broca's area have trouble speaking"). I would not techniques which correlate location and activity reverse engineering.

What I would classify as reverse engineering is pretty simple - electrophysiology. You can't get a much cleaner instance of "black boxin' " than ephys. Electrophysiology certainly predates Kurzweil's prediction (by decades). However, I wouldn't call it reverse engineering the brain. It's analogous to reverse engineering a transistor - certainly not reverse engineering an IC chip, but definitely a start.

However, reverse engineering entire neural circuits is a new field. From what I can tell, "schematics" of mouse brains (using dead mice) weren't mapped till the mid 2000s (2005-6). Furthermore, the process of relating circuit output with behavior output is even more recent. (mainly revolves around advances in genetic models that respond to light, and biomedical optics). I'd say that the process of 'black boxing' neural circuits for behavioral output is very recent (2009ish?).

There are many reasons for the major delays - 1) the process of getting structural data out of a brain wasn't viable until recently, 2) genetically engineering organisms to have inducible circuits used to be far more cumbersome, but is fairly streamlined now, 3) optical electrophysiology has gone a long way (i.e., lasers to fire neurons and lasers to measure neurons - or as Dr. Evil would say, 'lasers'), 4) My allegiance to RK demands that I self-censor anachronistic developments which fail to abide by his predictions. All hail RK!
"The Singularity Is Near" by Ray Kurzweil, How Close?? Quote
08-08-2010 , 02:38 AM
Quote:
Originally Posted by Plancer
I actually disagree here.
Quote:
Originally Posted by wiki
Reverse engineering is the process of discovering the technological principles of a device, object or system through analysis of its structure, function and operation.
I'd say they're both fair game. There's more than structure. And knowing density of white matter isn't exactly sufficient to make a black box either.
"The Singularity Is Near" by Ray Kurzweil, How Close?? Quote
08-08-2010 , 03:03 AM
Quote:
Originally Posted by Plancer
This interpretation of the prediction probably makes more sense when it was made than today. You call that interpretation "prima facie absurd" largely because the practice of customers purchasing completely digital objects has become mainstream.

This prediction was not obvious, and likely would have encountered a great deal of resistance / disbelief when presented ye olde simple folk.
Well, lots of predictions of technologies that actually exist today would probably wow the simple folk. Digital internet downloads, circa 1999, were not an aberration. alt.binaries.music had existed for years, p2p had existed for years, legitimate business offered legitimate mp3 downloads, a ton of software was available via (legitimate) download, movies were downloaded (on a smaller scale), etc. The question on internet downloads wasn't whether people's individual copies typically didn't come with associated physical objects- the question was whether any appreciable portion of them actually did.

In and before 1999 software and music were legally offered as internet downloads by real companies, and virtually none of them came with a physical object. You didn't buy an mp3 and get a CD mailed to you, or buy a CD and then do something to get a mp3 version. These were real businesses. mp3.com even ran a (successful) IPO in 1999 (after starting in 1997). This concept, digital-object-copy-only internet music downloads, both legitimate and pirate, predated the book. Internet-download-only software obviously massively predated the book- that's not even in question. Movies were probably only pirate at the time, and I don't know what the status of books was, and I'm tired and not going to do research right now.

Talking about digital-object-only internet music and software downloads as a future prediction is a ridiculous past-post in 1999, and it's basically impossible for him to have been unaware of the software side, and highly unlikely for him to have been unaware of the music or movie side (again, I don't know about books right now). As a non-past-post, his prediction can only mean that actual music albums exist as digital objects and stop having CDs associated with them- not that people's internet downloads don't come with CDs, because that was already the business model.
"The Singularity Is Near" by Ray Kurzweil, How Close?? Quote
08-08-2010 , 03:40 AM
Plancer is a wise one.
"The Singularity Is Near" by Ray Kurzweil, How Close?? Quote
08-08-2010 , 03:53 PM
Quote:
Originally Posted by Plancer
Computers are so stupid! Yesterday, I bumped into one that couldn't even solve NP complete problems...let alone break RSA!

I mean, com'on, who doesn't know how to solve NP complete problems?
I have no ide what the point of your post is. You just copied my correct post in a sarcastic tone. My post was correct because the claim was made that computers were good at solving (not just beating humans at) games. This is obv incorrect since computers can't even break RSA which is expected to be many orders of magnitude simpler than NP complete problems which in turn are expected to many orders of magnitude simpler than solving games.

EDIT: And I think all Kurzweil supporters ITT (and maybe even Kurzweil himself) should read up on computational complexity.

Last edited by Max Raker; 08-08-2010 at 04:05 PM.
"The Singularity Is Near" by Ray Kurzweil, How Close?? Quote
08-08-2010 , 04:22 PM
And combinatorial explosion.

Computers are SOOOO far away from understanding "ordinary language" which Kurzweil made a prediction about.
"The Singularity Is Near" by Ray Kurzweil, How Close?? Quote
08-08-2010 , 04:56 PM
Quote:
Originally Posted by Max Raker
I have no ide what the point of your post is. You just copied my correct post in a sarcastic tone. My post was correct because the claim was made that computers were good at solving (not just beating humans at) games. This is obv incorrect since computers can't even break RSA which is expected to be many orders of magnitude simpler than NP complete problems which in turn are expected to many orders of magnitude simpler than solving games.

EDIT: And I think all Kurzweil supporters ITT (and maybe even Kurzweil himself) should read up on computational complexity.
No, I agree with everything you said. My post certainly doesn't further any pro-Kurzweil argument.
I just thought it was funny reading "can't even" followed by some of the world's most difficult problems, and I figured I would never get a chance to make that joke again.
The post was meant purely in jest.
"The Singularity Is Near" by Ray Kurzweil, How Close?? Quote
08-08-2010 , 05:03 PM
Quote:
Originally Posted by Plancer
No, I agree with everything you said. My post certainly doesn't further any pro-Kurzweil argument.
I just thought it was funny reading "can't even" followed by some of the world's most difficult problems, and I figured I would never get a chance to make that joke again.
The post was meant purely in jest.
Oh ok. Sorry I misunderstood. You are correct, those problems ARE incredibly difficult. It seems like for Kurzweil to be correct, they need to all be solved very soon.
"The Singularity Is Near" by Ray Kurzweil, How Close?? Quote
08-08-2010 , 05:20 PM
I don't see how computational complexity has anything to do with the singularity.

I'd bet most RK supporters + RK are both versed in complexity theory and believe that better methods of generating heuristics will continue to emerge.
I've seen chess kicked around a few times in this thread - it's worth mentioning that chess is EXPTIME-complete, but that doesn't stop Fritz (a chess AI) from obliterating my ass. Chess AI doesn't operate via brute force - it actually is based on heuristics now (except for endgame calculations of course - but you don't get to do those till 20-40 moves deep, which is a long time in EXPTIME).

I don't really know how human civilization will be limited by computational complexity. I would guess there are major implications for economics / governance (which would favor a more libertarian stance), but I don't see this crossing over to brain / AI research. I would also guess that it would pose limits on bioengineering, but this is far-out science fiction stuff.
As far as we know, our brains never solve these problems - so why would an AI have to?
RK supporters don't think we're going to be able to crush minesweeper - we just expect godlike powers and immortality .
"The Singularity Is Near" by Ray Kurzweil, How Close?? Quote
08-08-2010 , 05:36 PM
Well, I think the chess example sometimes gets overstated. Computers have not solved chess anymore than Bobby Fischer solved it. They both got to a point where they were better than every (or every other in Fischer's case) human. That is a far cry from solving chess. Computational complexity seems relevant to me because the singularity seems so much harder than so many problems that we can't solve. Can we be post singularity, but somehow not be able to prove the Riemann Hypothesis or break a pretty low level (as far as computational complexity goes) encryption system like RSA? If these things are going to happen pre singularity (which I think they pretty much have to) they are going to have to fall pretty soon as the singularity itself seems like it will be many many orders of magnitude greater than these simple problems. I know the "answer" is something about an exponential..... but it is very hard for me to believe that such amazing stuff is possible in my lifetime, while such mundane stuff (that is 100% rigorously defined) cannot be done right now.
"The Singularity Is Near" by Ray Kurzweil, How Close?? Quote
08-08-2010 , 06:05 PM
I pretty much agree with everything you wrote, except I don't see why we would expect complexity to be conquered before the singularity.
Complexity is way more complicated than a single singularity.

I mentioned chess because it's an EXPTIME-complete problem which no one can solve, but the heuristics our computers use are so good that we can't beat them. I'd expect other intractable problems to go this route, whether it's playing a game of go or reading a few sentences of English.

I specifically said "RK supporters don't think we're going to be able to crush minesweeper - we just expect godlike powers and immortality" - I'm such a pesimist that I think it's plausible that we won't even have these answers postsingularity. Computation might just be screwed - it doesn't matter if the god-AIs can get 50 doublings in before they hit any theoretical limits, many many problems will be totally unsolvable. So, when I'm an ephemeral entity composed of information on a supercomputer, I will still get pwnd by Minesweeper (which is NP complete).

The RK thesis is that the singularity will be a byproduct of exponential growth. Of course, solutions to life / chess / go / RSA are DEFINED so that they can't be found with exponential growth.

I don't think there's any reason that the singularity will show up after major advances in our understanding of complexity. Furthermore, I don't know if it's even possible to understand complexity .
"The Singularity Is Near" by Ray Kurzweil, How Close?? Quote
08-08-2010 , 07:07 PM
Well, it is almost certainly possible to understand complexity. These are problems that are as well defined as anything else in math. And I don't really agree with you that immortality is probable, but somehow we can't do math problems that are certainly possible to solve. And RSA is likely much easier to than solving NP complete problems.
"The Singularity Is Near" by Ray Kurzweil, How Close?? Quote
08-08-2010 , 08:53 PM
Quote:
Originally Posted by Max Raker
Well, it is almost certainly possible to understand complexity. These are problems that are as well defined as anything else in math. And I don't really agree with you that immortality is probable, but somehow we can't do math problems that are certainly possible to solve. And RSA is likely much easier to than solving NP complete problems.
You are making the critical mistake of thinking that if a problem can be defined in a simple way then its solution should be simple.
"The Singularity Is Near" by Ray Kurzweil, How Close?? Quote
08-08-2010 , 10:26 PM
Quote:
Originally Posted by Karganeth
You are making the critical mistake of thinking that if a problem can be defined in a simple way then its solution should be simple.
Of course I'm not. I am talking about problems to which I could recognize the solution to that we can't do. Your post doesn't have anything to do with the conversation.
"The Singularity Is Near" by Ray Kurzweil, How Close?? Quote
08-08-2010 , 11:08 PM
Quote:
Originally Posted by Max Raker
Well, it is almost certainly possible to understand complexity. These are problems that are as well defined as anything else in math. And I don't really agree with you that immortality is probable, but somehow we can't do math problems that are certainly possible to solve. And RSA is likely much easier to than solving NP complete problems.
Well, we don't know if these math problems are solvable at all.
Almost everyone thinks NP != P, and a great deal of people think that you can't even prove it.

When SingularityTypes speak about immortality, we usually don't mean immortality in the sense that "all diseases will be cured by time x." Usually, immorality means "the average life expectancy will rise at a rate of at least one year, every year." RK has written extensively about this.

Naturally, predictions about life expectancy post-singularity don't make sense, because life post-singularity isn't supposed to make sense.

I think optimistic predictions regarding life expectancy are pretty reasonable, provided you add the caveat that you're discussing the life expectancies of wealthy first worlders.
"The Singularity Is Near" by Ray Kurzweil, How Close?? Quote

      
m