6 ms·

The Universe Is a Machine That Keeps Learning, Scientists Say

44 points by weatherlight 5y ago

bitwize 5y ago
> In fascinating new research, cosmologists explain the history of the universe as one of self-teaching, autodidactic algorithms. ISOs. Bio-digital jazz, man. The universe certainly contains machines that keep learning (lifeforms, and more broadly life itself) as part of itself, but I wonder how the stronger claim can be made, especially since there are parts of the universe being doomed to never interact with any other part as the universe expands and their local neighborhood is pushed beyond reach of particles travelling at c.
- uoaei 5y ago
  > especially since there are parts of the universe being doomed to never interact with any other part as the universe expands and their local neighborhood is pushed beyond reach of particles travelling at c It's not clear that this disqualifies the claim. Would you mind expanding? I am heavily biased toward what I call a "computationalist" point of view on metaphysics. This has resulted from the study of machine learning theory, being primarily Bayesian probability theory, statistical mechanics, computing theory, etc. and recognizing the deep isometries between them. This position seemingly inevitably leads one to a panbiotist stance, i.e., that there is no line but an arbitrary one between what can be considered "individual" vs "collective", nor between "living" and "non-living". In short, It is the Spinozan "God" for the Information Age. So I am in vigorous agreement with the conclusions of these cosmologists.
  - techbio 5y ago
    Also in vigorous agreement. Though not necessary to your points, I think GP's post is saying that omnipresent omniscience is not computable everywhere equally in the current best understanding of physics.
  - simorley 5y ago
    How can you believe in a "computationalist view of metaphysics" when computable numbers make up a tiny insignificant portion of real numbers? Most real numbers are not computable.
  - bitwize 5y ago
    The universe can't be a self-teaching machine if by virtue of its constant expansion it's breaking up into smaller and smaller pieces that are totally informationally isolated from one another. If anything, that means it's getting stupider.
gavinray 5y ago
This Popular Mechanics article sucks, as you no doubt guessed without clicking on it. How do these writers sleep at night. "Yeah, so I start my day -- I take the collective work of several brilliant people, and then I butcher it to fit in a few paragraphs between adspace and collect my paycheck. It's a good life." Gross. You know they can't have read the paper they're writing about. Because if they bothered to read the paper, it would take up the majority of their workday and they'd have no time to write anything. Also not sure how believable it is that someone academically talented enough to digest that paper is working as a writer for Pop Mechanics.
- tozze 5y ago
  Get back to work Gavin. Geez.
  - gavinray 5y ago
    It's Sunday mate.
- xwdv 5y ago
  Of course they don’t read it, they use an AI summarizer to pull out high scoring pieces of text then repackage it up to make an article out of it that gets fed to an end user for ad clicks.
lisper 5y ago
Here, let me save you some time: the paper that this article is about is a 79-page-long pseudo-scientific apologia for the fine-tuning argument [1], with the universe playing the role of God, but with the persona of a toddler playing around with a Lego set until it gets an arrangement of bricks that produces us because reasons. It's lipstick on a pig. Highly-credentialed lipstick, but still lipstick. Any serious defense of this idea needs to address one very basic question: what does this theory explain that the anthropic principle does not? On this question the paper is silent. There is one reference to the anthropic principle, and that is just to present the fine-tuning "problem". The paper says nothing else about it. So the entire thing is one giant exercise in question-begging [2]. [1] https://plato.stanford.edu/entries/fine-tuning/ https://plato.stanford.edu/entries/fine-tuning/ [2] https://www.txstate.edu/philosophy/resources/fallacy-definitions/Begging-the-Question.html https://www.txstate.edu/philosophy/resources/fallacy-definit...
phkahler 5y ago
I hope my tax dollars aren't funding this "research".
EMM_386 5y ago
It's important that minds (our consciousness) can be easily thought of as not part of the universe, but of course they are. Billions of minds on this planet alone (and rapidly increasing), all firmly part of the universe. So is the universe conscious? Of course it is. And it's learning about itself from all sorts of crazy angles.
- jstanley 5y ago
  > but of course they are. I don't think this is cut and dry. Consider a video game. Let's say Doom. All the characters inside Doom (including the player) exist within the Doom universe, but the mind controlling the player does not exist within the Doom universe. Yet there is no indication whatsoever, within the Doom universe, that there exists anything outside it. There is no (or, at least, not necessarily any) experiment you can do inside Doom to determine that the player's consciousness is outside Doom. We only know it is because we've seen outside. Why must our universe be different? We only assume our minds are within it because we haven't seen outside.
  - angelzen 5y ago
    Characters in Doom will soon discover that they have no plausible gradual explanation for the mind that controls the player. No matter how many fMRI scans of the player head they take, there is no signal.
    - jstanley 5y ago
      The fact that an fMRI scan shows a signal does not prove that the brain contains the mind.
      - robbedpeter 5y ago
        The fact that any state in which no fmri signal is present coincides 100% with a mind not being present is conclusive enough. Remove the neocortex and the mind is gone. Almost all of the rest of the brain's biology is optional to the phenomenon of mind, except in their respective roles of supporting the neocortex. Neuroscience can't really experiment the same way as other disciplines, but they can study injury and deformity and so-called lower animals, and some of the science is definitive. Unless you demand something metaphysical or spiritual or magical be involved in the human mind, in which case, why are you bothering with science at all? There is no magic. We are neurons relaying a dance of fascinating electrical patterns back and forth, somehow resulting in the construct of our subjective experiences. We have an answer to the question of the mechanics of consciousness. We don't have the algorithm yet, but we're making exponential progress in that direction as microscopy and biology increase the resolution of neural connectomes. The brain, specifically the neocortex, contains the mind, and there's no good reasons to assume otherwise.
delegate 5y ago
Worth noting that us humans are also Universe. So it hasn't only tried all laws of physics, it has written an article which it reads and comments on. Remembering that we're Universe observing itself, mathematics not only formulates the laws of 'physics' - it's Universe understanding and predicting itself.
Traubenfuchs 5y ago
What is it that decides what those laws are? Can it be changed, globally or within a confined space? Are „those laws“ the physical constants?
erdos4d 5y ago
This is pseudoscience.
bendbro 5y ago
Hi guys, Chief Accessibility Editor at Vice News here... feel like I need to call out why this title is a tad problematic. Nothing against the journalist that created it (we're all in this together) but it's important to stand up when teaching is needed. Titles are important- they tell the audience the value they will receive from a work- acting as the vector of the people through truth to power. Many people who read the news are working class, and because of this have a structural disadvantage that stunts formation of an attention span encouraging deep engagement with literature. To make news more equitable, it is incredibly important that titles pop, not just for the "typical" stodgy reader, but also for the common person, who's full potential can be unlocked by the world of learning woven in the fabric of well crafted words. So for this title I would have gone with something more like: "The Universe Is a Machine That Keeps Learning, Scientists Say, and That's a Good Thing! What this means in a post-trump america". This is not a criticism, its a teaching moment, and we'd do best to remember that.
ethn 5y ago
So any sane person will immediately recognize this sounds crazy. However, epistemologically it might be sound, I've been thinking about this for a while now. Mainly because of the simple potential well problem and considering Hegelian epistemology. The particle form isn't determined until it is measured, this is okay in regular epistemology and transcendental idealism. Not an issue, all appearances are human activities anyways created by our limited senses and tools. But the amazing thing is that the particle still complies with our concept of causality despite not yet existing (its probabilities are completely dependent on potential energy)! There are several theories to try to reconcile this. Real many-worlds theory, holographic universe, superdeterminism, single-electron etc. But all of these have either epistemological issues or cannot account for all phenomena. This has led me to what I think is the most satisfying interpretation so far, which I call retroactive superdeterminism (RS). In RS in order to resolve the problem of maintaining causality for an object which does not exist, on information of the appearance of the particle which may otherwise contradict causality, the mind changes the entire history of the world so to be that the particle existed the entire time with defined position and momentum, thereby producing causality. Thereby saving the mental framework we have on all things (phenomenology). In this way the universe 'learns' but of course it is merely the ensemble of the appearances of the world and their properties that are changing as a result of that mental activity, which from one perspective can be argued that the universe is learning.
jshaqaw 5y ago
Some of the mathematical equivalences worked out on the underlying article are quite interesting. The broader implications part is fairly hand-wavy vague but you can skip it. Part of me wonders if Smolin is trolling the string theorists with “well this idea is mathematically consistent if untestable too…”
Grismar 5y ago
"Scientists say" - I just hope people realise that, in cases like these, that means "a scientist or two wrote about their opinion", which is not even related to "there is scientific consensus". Also, sometimes scientists are really bad at their job.
hulitu 5y ago
We all must understand that the universe is only about AI, ML, UX, Blockchain and NFT. We are only slowly been able to unleash the hidden power of cryptocurrencies.

44 points by weatherlight 5y ago

> In fascinating new research, cosmologists explain the history of the universe as one of self-teaching, autodidactic algorithms. ISOs. Bio-digital jazz, man. The universe certainly contains machines that keep learning (lifeforms, and more broadly life itself) as part of itself, but I wonder how the stronger claim can be made, especially since there are parts of the universe being doomed to never interact with any other part as the universe expands and their local neighborhood is pushed beyond reach of particles travelling at c.

uoaei 5y ago
> especially since there are parts of the universe being doomed to never interact with any other part as the universe expands and their local neighborhood is pushed beyond reach of particles travelling at c It's not clear that this disqualifies the claim. Would you mind expanding? I am heavily biased toward what I call a "computationalist" point of view on metaphysics. This has resulted from the study of machine learning theory, being primarily Bayesian probability theory, statistical mechanics, computing theory, etc. and recognizing the deep isometries between them. This position seemingly inevitably leads one to a panbiotist stance, i.e., that there is no line but an arbitrary one between what can be considered "individual" vs "collective", nor between "living" and "non-living". In short, It is the Spinozan "God" for the Information Age. So I am in vigorous agreement with the conclusions of these cosmologists.
- techbio 5y ago
  Also in vigorous agreement. Though not necessary to your points, I think GP's post is saying that omnipresent omniscience is not computable everywhere equally in the current best understanding of physics.
- simorley 5y ago
  How can you believe in a "computationalist view of metaphysics" when computable numbers make up a tiny insignificant portion of real numbers? Most real numbers are not computable.
- bitwize 5y ago
  The universe can't be a self-teaching machine if by virtue of its constant expansion it's breaking up into smaller and smaller pieces that are totally informationally isolated from one another. If anything, that means it's getting stupider.

gavinray 5y ago

This Popular Mechanics article sucks, as you no doubt guessed without clicking on it. How do these writers sleep at night. "Yeah, so I start my day -- I take the collective work of several brilliant people, and then I butcher it to fit in a few paragraphs between adspace and collect my paycheck. It's a good life." Gross. You know they can't have read the paper they're writing about. Because if they bothered to read the paper, it would take up the majority of their workday and they'd have no time to write anything. Also not sure how believable it is that someone academically talented enough to digest that paper is working as a writer for Pop Mechanics.

tozze 5y ago
Get back to work Gavin. Geez.
- gavinray 5y ago
  It's Sunday mate.
xwdv 5y ago
Of course they don’t read it, they use an AI summarizer to pull out high scoring pieces of text then repackage it up to make an article out of it that gets fed to an end user for ad clicks.

lisper 5y ago

Here, let me save you some time: the paper that this article is about is a 79-page-long pseudo-scientific apologia for the fine-tuning argument [1], with the universe playing the role of God, but with the persona of a toddler playing around with a Lego set until it gets an arrangement of bricks that produces us because reasons. It's lipstick on a pig. Highly-credentialed lipstick, but still lipstick. Any serious defense of this idea needs to address one very basic question: what does this theory explain that the anthropic principle does not? On this question the paper is silent. There is one reference to the anthropic principle, and that is just to present the fine-tuning "problem". The paper says nothing else about it. So the entire thing is one giant exercise in question-begging [2]. [1] https://plato.stanford.edu/entries/fine-tuning/ https://plato.stanford.edu/entries/fine-tuning/ [2] https://www.txstate.edu/philosophy/resources/fallacy-definitions/Begging-the-Question.html https://www.txstate.edu/philosophy/resources/fallacy-definit...

phkahler 5y ago

I hope my tax dollars aren't funding this "research".

EMM_386 5y ago

It's important that minds (our consciousness) can be easily thought of as not part of the universe, but of course they are. Billions of minds on this planet alone (and rapidly increasing), all firmly part of the universe. So is the universe conscious? Of course it is. And it's learning about itself from all sorts of crazy angles.

jstanley 5y ago
> but of course they are. I don't think this is cut and dry. Consider a video game. Let's say Doom. All the characters inside Doom (including the player) exist within the Doom universe, but the mind controlling the player does not exist within the Doom universe. Yet there is no indication whatsoever, within the Doom universe, that there exists anything outside it. There is no (or, at least, not necessarily any) experiment you can do inside Doom to determine that the player's consciousness is outside Doom. We only know it is because we've seen outside. Why must our universe be different? We only assume our minds are within it because we haven't seen outside.
- angelzen 5y ago
  Characters in Doom will soon discover that they have no plausible gradual explanation for the mind that controls the player. No matter how many fMRI scans of the player head they take, there is no signal.
  - jstanley 5y ago
    The fact that an fMRI scan shows a signal does not prove that the brain contains the mind.
    - robbedpeter 5y ago
      The fact that any state in which no fmri signal is present coincides 100% with a mind not being present is conclusive enough. Remove the neocortex and the mind is gone. Almost all of the rest of the brain's biology is optional to the phenomenon of mind, except in their respective roles of supporting the neocortex. Neuroscience can't really experiment the same way as other disciplines, but they can study injury and deformity and so-called lower animals, and some of the science is definitive. Unless you demand something metaphysical or spiritual or magical be involved in the human mind, in which case, why are you bothering with science at all? There is no magic. We are neurons relaying a dance of fascinating electrical patterns back and forth, somehow resulting in the construct of our subjective experiences. We have an answer to the question of the mechanics of consciousness. We don't have the algorithm yet, but we're making exponential progress in that direction as microscopy and biology increase the resolution of neural connectomes. The brain, specifically the neocortex, contains the mind, and there's no good reasons to assume otherwise.

delegate 5y ago

Worth noting that us humans are also Universe. So it hasn't only tried all laws of physics, it has written an article which it reads and comments on. Remembering that we're Universe observing itself, mathematics not only formulates the laws of 'physics' - it's Universe understanding and predicting itself.

Traubenfuchs 5y ago

What is it that decides what those laws are? Can it be changed, globally or within a confined space? Are „those laws“ the physical constants?

erdos4d 5y ago

This is pseudoscience.

bendbro 5y ago

Hi guys, Chief Accessibility Editor at Vice News here... feel like I need to call out why this title is a tad problematic. Nothing against the journalist that created it (we're all in this together) but it's important to stand up when teaching is needed. Titles are important- they tell the audience the value they will receive from a work- acting as the vector of the people through truth to power. Many people who read the news are working class, and because of this have a structural disadvantage that stunts formation of an attention span encouraging deep engagement with literature. To make news more equitable, it is incredibly important that titles pop, not just for the "typical" stodgy reader, but also for the common person, who's full potential can be unlocked by the world of learning woven in the fabric of well crafted words. So for this title I would have gone with something more like: "The Universe Is a Machine That Keeps Learning, Scientists Say, and That's a Good Thing! What this means in a post-trump america". This is not a criticism, its a teaching moment, and we'd do best to remember that.

ethn 5y ago

So any sane person will immediately recognize this sounds crazy. However, epistemologically it might be sound, I've been thinking about this for a while now. Mainly because of the simple potential well problem and considering Hegelian epistemology. The particle form isn't determined until it is measured, this is okay in regular epistemology and transcendental idealism. Not an issue, all appearances are human activities anyways created by our limited senses and tools. But the amazing thing is that the particle still complies with our concept of causality despite not yet existing (its probabilities are completely dependent on potential energy)! There are several theories to try to reconcile this. Real many-worlds theory, holographic universe, superdeterminism, single-electron etc. But all of these have either epistemological issues or cannot account for all phenomena. This has led me to what I think is the most satisfying interpretation so far, which I call retroactive superdeterminism (RS). In RS in order to resolve the problem of maintaining causality for an object which does not exist, on information of the appearance of the particle which may otherwise contradict causality, the mind changes the entire history of the world so to be that the particle existed the entire time with defined position and momentum, thereby producing causality. Thereby saving the mental framework we have on all things (phenomenology). In this way the universe 'learns' but of course it is merely the ensemble of the appearances of the world and their properties that are changing as a result of that mental activity, which from one perspective can be argued that the universe is learning.

jshaqaw 5y ago

Some of the mathematical equivalences worked out on the underlying article are quite interesting. The broader implications part is fairly hand-wavy vague but you can skip it. Part of me wonders if Smolin is trolling the string theorists with “well this idea is mathematically consistent if untestable too…”

Grismar 5y ago

"Scientists say" - I just hope people realise that, in cases like these, that means "a scientist or two wrote about their opinion", which is not even related to "there is scientific consensus". Also, sometimes scientists are really bad at their job.

hulitu 5y ago

We all must understand that the universe is only about AI, ML, UX, Blockchain and NFT. We are only slowly been able to unleash the hidden power of cryptocurrencies.