Whoa, dude, are we in a computer right now? NASA scientist thinks we could be.
vice.com
vice.com
Wow this was a really good article. I didn't realize that we are
so close to simulating consciousness. and this part: "In quantum
mechanics, particles do not have a definite state unless they’re
being observed." is that true? That blew my mind.
First of all, there is a whole spectrum of simulation software to consider. The most Matrix-like method is a direct 3D game-like environment with really good graphics where many parameters of our current environment are hard-coded and everything is executed with the human inhabitant in mind. This is probably close to what the article is trying to invoke if for no other reason than "it's easy to understand". It's also the one which least conforms to our observations. If the universe is being simulated, it's more likely a bottom-up simulation where things like planets, plants, and eventually people simply emerge instead of being designed outright. For all intents and purposes, such a universe would be indistinguishable from a "natural" one, so much so it's bordering on pointlessness even arguing about it.
Second, simulating consciousness (or rather more precisely: designing and running human-like minds on a digital substrate) directly is not something we're particularly close to. We are still missing important engineering details about human brains to recreate them, and likewise generalised AGI research also has some gaps in understanding how to build advanced problem solving machines. Depending on the chosen method of "simulation", from full no-shortcuts whole brain emulation probably being the most expensive to fast and memory-optimised non-human AGI being the more efficient, we might already have enough raw computing power to drive a mind using just a couple of desktop computers. What we're missing though is the right software.
I'm also getting really tired about this whole "observer" misunderstanding whenever quantum physics is invoked. I can't help but feel this was an intentional consequence when they chose their terminology. To this day it remains a poor choice of wording. The agenda behind this might not only be designed to sell books to New Age lunatics, but also to enable pseudo-religious excursions undertaken by otherwise respectable scientists such as Penrose.
Well, yes, I do actually. See philosophical zombie (http://en.wikipedia.org/wiki/Philosophical_zombie). Although I don't doubt that the physics of our universe could be simulated.
By the way, there's actually something called the Bekenstein bound (http://en.wikipedia.org/wiki/Bekenstein_bound) which puts a finite information limit on a volume of spacetime. Incredibly enough, the information content is not proportional to the volume of the region, but to the volume's surface area. This is known as the holographic principle (http://en.wikipedia.org/wiki/Holographic_principle).
I did a calculation the other day for fun with the Bekenstein bound. All of the information in 1 kg of water can be entirely described with roughly ~20,000 earth masses of 1 TB hard drives (assuming each hard drive weighs one pound).
And if someone wants to see something extra freaky, calculate the maximal information content of a Planck volume ;)
What constitutes science? You must be able to predict events from a theory and you must be able to reproduce them. But there's a third, implied tenet that most people forget about: anyone performing the experiment must come to a consensus on the results of the experiment.
For me, "experience" is easy -- it's the most "real" thing in my reality. Vision, hearing, taste, smell? Those could all be faked some day with advanced enough neural manipulation.
Let me attempt to clarify.
Suppose you have a hypothetical person who sees an apple floating everywhere in front of him. He tells everyone that he sees this apple, but everyone else says "No, you're crazy. There is no apple there." So he takes a photograph of the apple. He prints it out and says "Look! Right here in this photograph is the apple!" Again, everyone says "I see no apple." He can consistently, predictably, see this apple. He decides to even paint exactly what he sees. And when he looks at his painting, he sees an apple, but everyone else sees empty sky.
Is this person crazy? Well, if science is determined by consensus (which it is), then yes, the majority of people would say that he is crazy. But now instead of just this one guy with an apple, let's say that everyone sees a piece of fruit in front of them. But it's a different type of fruit for each person. And they all try to tell each other that there is clearly, obviously, unmistakably a fruit floating right there in front of them, but everyone thinks everyone else is crazy because they can't see other people's fruit -- only their own.
So eventually, they decide to just forget about the fruit. Maybe it's not actually real if everyone else can't see it? (Would a person in a computer simulation see this fruit?) And so, despite everyone obviously having their own fruit hanging right in front of them, they have all decided to say that there is instead no such thing as floating fruit, since no one can come to a consensus about everyone else's fruit existing.
So there you go. That's how I view consciousness. Although I should tweak that slightly -- "consciousness" perhaps isn't the right word. Consciousness can be manipulated with drugs and sedatives. What I am really referring to is the act of "experiencing" something.
First: 'Science' does not embody the whole of human knowledge, not at all. That's not even what it's for. It is a true thing that 1+1=2 (with standard definitions, Peano arithmetic, etc), but science has nothing to say on this matter.
Second: The fact that other people talk about this fruit they see, and you yourself have this fruit that you see, is very strong evidence that people have fruits, even if, with our current understanding, we can't verify it from physics.
Third: I do believe consciousness exists. I think (in fact it seems self-apparent to me) that consciousness is a property of processes or algorithms. Can I explain how exactly it arises? No, but then for virtually all of human history we had no idea how, say, biology or magnets or fire could come out of physics either.
Fourth: Just because a process has some property doesn't mean we'll ever understand it in a satisfying way - even very simple processes (care to explain how the current best-candidate for BusyBeaver(6) works?) This is not the same as that property not existing.
Finally: This was in my original post as well, was in fact the entire point of said post, and I can't seem to find a response in your post. If you accept (as is a part of the p-zombie postulation) that the reasons you say things like 'I am conscious' are entirely because of physics, do you think that this means your consciousness has nothing to do with your claim to be conscious? That seems absurd to me, and to me it seems that the conclusion must be, given the above, that consciousness does in fact fall out of physics.
For clarity, here's that final argument spelled out again.
1. My claim to be conscious does have to do with the fact of my consciousness. (Agree?)
2. My claim to be conscious is completely determined by physics, since I would also make it in p-zombie world (that is, the world which is by hypothesis completely determined by physics). (Agree?)
3. Therefore, the fact of my consciousness must fall out of physics. (Agree?)
Or in set-theoretic terms, because that's how I think:
fact-of-consciousness \subset reasons-for-claim-to-be-conscious \subset physics implies fact-of-consciousness \subset physics.
Do you disagree?
First: Yes, I agree.
Second: Hmm... I don't know. People talk about ghosts a lot as well. There are many people who claim that they see ghosts. But I don't think that is strong evidence that ghosts exist.
Third: We need a consistent definition of consciousness here. It would waste a lot of time if it turned out we were debating different concepts. I believe your idea of consciousness is that it is an emergent phenomenon that arises from simpler processes, much like how biology arises from the simple laws of QM. Is this correct? If so, then with that definition of consciousness, I agree with you. And we will learn more about it as time goes on.
Fourth: Yes, I agree. Chaitin's constant is my favorite exemplification of that idea.
Finally: "Do you think that this means your consciousness has nothing to do with your claim to be conscious?" Yes, that is exactly what I think. Although replace "consciousness" with "the act of experiencing existence" and that is a much closer semantic match to what I am attempting to convey.
> 1. My claim to be conscious does have to do with the fact of my consciousness. (Agree?)
No, for the simple reason that you can write a computer program that prints out "I am conscious." Its claim of consciousness is only due to a program that someone wrote.
> 2. My claim to be conscious is completely determined by physics, since I would also make it in p-zombie world. (Agree?)
Yes, I agree with that.
> 3. Therefore, the fact of my consciousness must fall out of physics.
Were 1 and 2 true statements, they would indeed imply 3. But since I don't agree with 1, 3 also doesn't seem to be true.
I wouldn't call that a definition. Here's a working one: my consciousness is the part of me which thinks and feels, which enjoys music and mountains and friends and so forth. Computers as of now can take input, but can't experience input the way I do; the gap is what I call consciousness. (To this definition your above description does apply.)
> "Do you think that this means your consciousness has nothing to do with your claim to be conscious?" Yes, that is exactly what I think. Although replace "consciousness" with "the act of experiencing existence" and that is a much closer semantic match to what I am attempting to convey.
This is the bit I'm having trouble with. For me, I experience things, and then state that I experience things. The relationship is direct and causal. From what I understand you experience things and then state that you experience things, and these facts are completely unrelated? I'm pretty sure Occam had something to say about that.
"Causation from particle-wave interactions". What do you mean by this? If you're referring to determinism, that's already been ruled out by HUP (unless you support superdeterminism, although few scientists do; t'Hooft is a notable exception).
Valid data type? When did the discussion switch to type systems? I'm sorry, you'll need to clarify what you are trying to convey.
I'm sorry your highness. I didn't realize I'm speaking to an authority here.
>Causation from particle-wave interactions". What do you mean by this?
I mean saying firing of neural circuitry X correlates to experience of Y. Given the right tools you might be able to do that and create a valid mapping of causation.
>Valid data type? When did the discussion switch to type systems?
What you can't do is capture experience in formal language. We'd like to symbolically capture the experience of "seeing blue" for example but no data type maps onto qualities beyond nominal signifiers of our own personal experience.
An easy example of this would be some recent work done in using MRI to map active brain regions to moral dilemmas. (A quick example: http://www.wbur.org/npr/6636701/the-science-behind-making-mo...)
To respond directly to your hypothetical apple, there are so many myriad ways of testing the existence or not-existence of a thing that hand-waving science away as agreement-by-consensus is premature (and wrong). Your hypothetical crazy dude could try immersing his apple in water and see if there is displacement; he could try burning it, measuring the light of the flame using a spectrometer, ask others to burn apples in the same way, and compare the output; he could attempt to infect it with insect larvae; he could attempt to power a small device with it; and so on. Science is not merely a matter of consensus, it is a matter of ever-rigorous testing.
Or, to put it another way: the closest thing that scientists have found so far to a "consensus" reality is relativity, and they found a way to test and verify it. If scientists can come up with a way to test whether or not time itself actually speeds up or slows down according to frame of reference, don't you think they could come up with a way to test for the existence of a magic invisible apple?
To get back to the article, I have to defer to Descartes: I think, therefore I am. And, I have thought about such things as MRI machines that verify the consciousness of the beings around me; their brains, the existence of their consciousnesses, is no longer a magical thing to me. I suppose it is possible that I am an element of some kind of incredible extra-dimensional simulation, but that's a very boring thing to imagine, because it doesn't in any way affect who I am or what I will do, or have done.
And besides, I think that one day scientists will be able to test that, too.
edit: here, I read the WP article you linked more thoroughly, so I'll just reference it back: I believe I would fall into the reductive physicalist camp, where p-zombies are impossible by definition. Furthermore, I think that the relatively recent work done in neuroscience -- work more recent than the idea of p-zombies, I would point out -- provides strong support for reductive physicalism.
P-zombie's are hypothetical people that lack a certain "property" (for lack of a better word) that other people have. This property is only testable on an individual basis. Since this property is not reproducible for agreement by consensus, you're right -- it isn't science. But it was never a purely scientific discussion to begin with anyway as we were debating the nature of consciousness/experience; no scientist does this as part of their job.
If you are referring to "consciousness" as a specific pattern of neural activity that occurs within the brain that can be studied, and from which additional correlations and predictions can be made, then yes, of course we are learning about that. We may even be able to simulate a brain in its entirety one day (heck, my research is in simulation so I can certainly see where it's headed [hundreds of years down the road I might add]). I don't think anyone on HN would dispute the improving science of neural analysis.
I've been reading a lot of David Chalmers lately - he's awesome.
The paper is very readable, even for the novice.
tl;dr:
"A technologically mature “posthuman” civilization would have enormous computing power. Based on this empirical fact, the simulation argument shows that at least one of the following propositions is true: (1) The fraction of human-level civilizations that reach a posthuman stage is very close to zero; (2) The fraction of posthuman civilizations that are interested in running ancestor-simulations is very close to zero; (3) The fraction of all people with our kind of experiences that are living in a simulation is very close to one. If (1) is true, then we will almost certainly go extinct before reaching posthumanity. If (2) is true, then there must be a strong convergence among the courses of advanced civilizations so that virtually none contains any relatively wealthy individuals who desire to run ancestor-simulations and are free to do so. If (3) is true, then we almost certainly live in a simulation. In the dark forest of our current ignorance, it seems sensible to apportion one’s credence roughly evenly between (1), (2), and (3). Unless we are now living in a simulation, our descendants will almost certainly never run an ancestor-simulation."
It wouldn't make us, or our experiences, any less real. The environment we evolved in would always be exceptionally realistic to us, by definition (we don't know any different!). Even if some higher entity sees us living on some game-of-life grid. Though that's a philosophical matter, I guess.
Edit: To be clear, I believe it is possible that we live in a matter-level simulation. I do not believe that we live in a human-level simulation. That would be a much too anthropocentric view for me.
Luckily, in Real Life I'm a dualist.
There are plenty of reasons. First, it's much easier to just copy a known world than create a complete new physics and beings. Just that, creating s coherent set of rules governing the universe would be a major undertaking. And I'm not even convinced that there are multiple types of intelligence, ie. I think all intelligent beings are fundamentally similar to each other.
Secondly, simulations similar enough seem to be more entertaining. Thirdly, mostly simulations similar enough would be useful as experiments.
As to #2, I'm not sure I agree. We've got lots of computer environments ourselves that aren't very lifelike at all, from Conway's Life to Sim-(fill-in-the-blank) to World of Warcraft to extremely specialized (and non-entertaining) environments. To some extent, that's because we don't have the processing power to simulate real life, but even with infinite processing power, I would guess that most simulations would not be perfectly lifelike. There are more lifelike simulations available today, but I don't think they're a majority.
For #3, yes, but maybe you're assuming the main purpose of simulation would be for experiments, and therefore that they're similar to the real world. From what I've seen of my world, the most common use of computers is entertainment, and for this, being different from reality (either a little or a lot) is very important. Research will have human brains-in-a-computer first, but a decade or two later the Playstation 7 will have a lot more of them. It's like asking: you're a master-level chess computer, who are you? You might be Deep Blue in 1996, but you're far more likely to be a chess program running on somebody's PC a few years later.
[0] I'm tempted to say this is most likely, but I don't think we even know enough to assign meaningful probabilities. We don't have any idea what the space of possibilities is.
"If you make a simple calculation using Moore’s Law, you’ll find that these supercomputers, inside of a decade, will have the ability to compute an entire human lifetime of 80 years—including every thought ever conceived during that lifetime—in the span of a month."
And said to myself "Bzzzzzt! and thanks for playing." But I expect we'll see more before we see less, folks like my kids are graduating from college and they have had access to computers their "whole life" and they have always managed to get hugely faster and more powerful "their whole life" and so it becomes something of an assumption.
When Intel drove the Pentium 4 off the process rails into the valley of heat death a lot of people said "Hmmm, that doesn't bode well." But the truth was even then memory speeds were flattening out, actual 'mean time to data' disk speeds have been slowing in lock step with the increase in density, and transactional state space that point in time where you know a transaction is well and truly complete, has gotten fuzzier and fuzzier.
So while I enjoy the simulation hypothesis, and remember going to my friends with a short story based on it (the punch line was once you figured out you were in a simulation it crashed, which caused the people running that simulation to realize it and crash that one, all they way up to a programmer who sighs, and says "Ok, lets start again, first we need some light ..." I thought it was hugely clever and was quite proud of it, the editor at Analog (magazine I sent it to) just wrote "simulation hypothesis, no thanks." on the post card I got back. Well suffice it to say its not all that new a concept.
As wikipedia nicely puts it:
Scientific laws:
1) summarize a large collection of facts determined by experiment into a single statement,
2) can usually be formulated mathematically as one or several statements or equation, or at least stated in a single sentence, so that it can be used to predict the outcome of an experiment, given the initial, boundary, and other physical conditions of the processes which take place,
3) are strongly supported by empirical evidence - they are scientific knowledge that experiments have repeatedly verified (and never falsified). Their accuracy does not change when new theories are worked out, but rather the scope of application, since the equation (if any) representing the law does not change. As with other scientific knowledge, they do not have absolute certainty like mathematical theorems or identities, and it is always possible for a law to be overturned by future observations.
"Law" is not a stronger form of "theory", as is popularly believed.
Now granted, it is not a particularly good law. Strictly speaking it is bad science, but a law it is nevertheless.
i.e., so far there seems to be no support for the idea that the universe is actually chaotic in the mathematical sense from the quantum level to the relativistic one; Brownian motion doesn't change the weather.
(I'm speaking specifically about the laws of physics in our geological era; the physics a few microseconds after the Big Bang would likely be different in this regard.)
"Quantum physics might seem to undermine the idea that nature is governed by laws, but that is not the case. Instead it leads us to accept a new form of determinism: Given the state of a system at some time, the laws of nature determine the probabilities of various futures and pasts rather than determining the future and past with certainty."
It's an interesting idea, because to me it suggests that if one were to rewind time and play events out again without changing anything, subtle things may occur differently. This would lead to larger time changes over a larger period of time. The idea seems strange, but then, so do many things on the quantum level as they are completely outside our traditional experiences.
Apparently no consensus has been reached on this topic by experts in the field. I won't try and convince you, just present my interpretation.
Interesting but untestable, and has no effect on our scientific understanding of the world. Fun thought to have but ultimately useless.
This hypothesis—versions of which have been kicked around for centuries—is becoming the trippy notion of the moment for philosophers, with people like Nick Bostrom, the director of Oxford University’s Future of Humanity Institute, seriously considering the premise.
Are we really sure it's untestable? It seems like something that might be worth thinking about. If we look in the right places (theoretical math, physics?) might we be able to see reflections of a world "outside"?
I was talking about this bit, which is a bit lower:
"but he is one of the first to argue we might already be living inside one."
>>Are we really sure it's untestable? It seems like something that might be worth thinking about. If we look in the right places (theoretical math, physics?) might we be able to see reflections of a world "outside"?
We can never be 100% sure it's untestable because it's impossible to prove a negative. But I put it in the same basket as testing the existance of a god.
Just like a system that takes some energy in cannot output more energy than what it took in, I think a simulation that takes some computational power in cannot output more computation than the original system is capable of.
If we had a simulation of the relevant aspects of the universe, of the mind, of computers, etc. that was capable of running another simulation inside it, in order to run the other simulation while still running itself it would have to double its computer power (or half its simulation speed), it would have to share computational power with any simulation running inside it, whether directly or nested inside another one. Without creating more computer power out of nothing it would soon come to the physical limits (or the computational limits) and would not be able to go further.
Then androids could simulate the combination of DNA from parents and have baby androids. :o
well that is the fun bit. :)
(If you've played Minecraft, a world simulator, you know you need a seed number to get the world started. It's the "let there be light" and ensures your Minecraft world is unique.)
So no, we aren't.