I am both excited and scared by this. I don't know why. I think maybe I've read too much dystopian fiction. What are the implications of this? What are the possibilities for this technology?
I am both excited and scared by this. I don't know why. I think maybe I've read too much dystopian fiction.
What are the implications of this? What are the possibilities for this technology?
The big thing I can think of is that silicone, by itself, can't think it can only process. We (humans) have come up with clever tricks to emulate thinking, but (assuming this does what's claimed)...
The big thing I can think of is that silicone, by itself, can't think it can only process. We (humans) have come up with clever tricks to emulate thinking, but (assuming this does what's claimed) this would truly allow computers to think like humans. The implications of that are pretty far-reaching. Think self-driving vehicles or robotics. The SUPER interesting bit to me is this combination of traditional silicone with brain tissue here. It makes me wonder if we can have the best of both worlds with an an AI that can truly think with a computer that can logically make the best decision. Not an expert in any related field (unless you count IT) so those are more my musing than anything else, lol.
Silicon, not silicone. Not that that's super important - just a common typo :D I don't think it's fair to say silicon can only process. What if we had a computer powerful enough to simulate a...
Silicon, not silicone. Not that that's super important - just a common typo :D
I don't think it's fair to say silicon can only process. What if we had a computer powerful enough to simulate a large amount of matter? And then have the computer simulate a full human body and brain. Would it not be thinking?
Or maybe we have a computer simulate all of the neurons in a brain. Is it thinking?
Or what if our non-biologically designed software neural networks get powerful enough to do all of the things a human mind can. Is it thinking?
No, it wouldn't. The analogy I like is this: Imagine you have 2 people who have learned to give the answers to multiplication problems. Alice learned to do column multiplication. Bob has memorized...
Would it not be thinking?
No, it wouldn't. The analogy I like is this: Imagine you have 2 people who have learned to give the answers to multiplication problems. Alice learned to do column multiplication. Bob has memorized a multiplication table all the way out to 100,000 x 100,000. Most of the time, it'll be impossible to detect the difference between Alice and Bob solving (though Bob might be a lot faster). But Bob is not really multiplying, whereas Alice is. And to further the analogy, we can make Bob's simulation of multiplying even better, by teaching him all the way to 1,000,000 (Bob has a very good memory). Yet still only Alice is capable of true multiplication, able to multiply any numbers we give her. But Bob is an excellent approximation, impossible to tell the difference most of the time.
I don't think that's a good analogy. In teaearlgraycold's hypothetical, the simulated brain would be doing exactly the same things the biological brain is doing. I see no difference between a real...
I don't think that's a good analogy. In teaearlgraycold's hypothetical, the simulated brain would be doing exactly the same things the biological brain is doing.
I see no difference between a real wet human brain and a perfectly simulated silicon one. As far as we know, consciousness is an emergent property of the complex systems in our brains, and if a computer could perfectly simulate those systems, then there's no reason to say that the consciousness that emerges is any less legitimate. The only objection to that would be that humans can think because we have souls, which as a naturalist, I don't believe.
Ok, that's fair. I guess I'm approaching from the POV that this hypothetical seems incredibly unrealistic, because every attempt we've made to model so far doesn't look anything like what humans...
Ok, that's fair. I guess I'm approaching from the POV that this hypothetical seems incredibly unrealistic, because every attempt we've made to model so far doesn't look anything like what humans do, and to a large extent we don't even know what humans do enough to be able to intentionally model it (to the best of my knowledge at least). In fact I consider it somewhat of a paradox if we were able to intentionally model true thinking, because that means there's a 0% chance that we're not intentionally modeled by a higher-order universe, with infinitely many higher-order universes above.
I feel like I'm getting into the weeds here, but the fact that we can't currently simulate a brain doesn't mean it isn't possible; it only means that we can't do it yet. While the inclusion of...
I feel like I'm getting into the weeds here, but the fact that we can't currently simulate a brain doesn't mean it isn't possible; it only means that we can't do it yet. While the inclusion of brain cells into electronics could speed up the development of truly "thinking" machines, I was simply agreeing with and defending teaearlgraycold's statement that it's a bit unfair to say that a computer made with brain cells can do things that silicon fundamentally cannot.
if we were able to intentionally model true thinking, because that means there's a 0% chance that we're not intentionally modeled by a higher-order universe, with infinitely many higher-order universes above.
I don't see how this follows. That we can intentionally model true thinking doesn't say anything about our own development as biological beings. We were still shaped by billions of years of evolution, and the development of intelligent synthetic life wouldn't change that.
I would like to throw at this discussion the fact that we don't understand the brain 100%, and we can't model something if we don't understand how it works. We can't even tell right now if it's...
I would like to throw at this discussion the fact that we don't understand the brain 100%, and we can't model something if we don't understand how it works. We can't even tell right now if it's possible with the current limitations in Computer Science, because we don't understand what we want to simulate, so we don't know what we need to simulate it correctly.
So, it could be possible, but there's also the possibility that we can't with our technology, and to be able to really do it, before we build an even better computer we have to understand how we think.
One example is about how language makes us able to think, what's the relationship between language, culture and individual thinking, what would happen to a human that didn't learn any language whatsoever, and so on.
We can't copy from scratch what we can't understand.
It would be more simulation than modeling. Modeling would mean an approximation. We’d probably model the cells but copy the connections from a brain scan.
It would be more simulation than modeling. Modeling would mean an approximation. We’d probably model the cells but copy the connections from a brain scan.
I think that’s a good argument for why a simulated human is not a real human. But not why a machine can not think - unless we define thinking as exclusively what a real human brain does.
I think that’s a good argument for why a simulated human is not a real human. But not why a machine can not think - unless we define thinking as exclusively what a real human brain does.
If a simulation is functionally equivalent to "the real thing," then what's the difference? It doesn't even really make sense to make a philosophical distinction, as philosophy divorced from...
If a simulation is functionally equivalent to "the real thing," then what's the difference? It doesn't even really make sense to make a philosophical distinction, as philosophy divorced from external reality is just navel-gazing debate about angels boogieing down on pinheads.
We don't know exactly what gives rise to sapience, though I think it likely to be an emergent phenomenon of complex intra-acting systems. If that's true, the difference between a human brain and a silicon semiconductor-based thinking machine is only a matter of degree, not kind.
I do wonder how far you can push this analogy. Take the xkcd "A Bunch of Rocks". If you could create a simulation of the universe just using a limitless supply of rocks, does that mean you could...
I do wonder how far you can push this analogy. Take the xkcd "A Bunch of Rocks". If you could create a simulation of the universe just using a limitless supply of rocks, does that mean you could simulate consciousness too? If so, where would the consciousness reside, so to speak?
Personally in the camp of materialism being a lower bounds. Matter structured and changing in a particular way appears to give rise to consciousness. Neurologists could see a brain scan of someone...
Personally in the camp of materialism being a lower bounds. Matter structured and changing in a particular way appears to give rise to consciousness. Neurologists could see a brain scan of someone with some particularly well-documented damage and infer something about their qualia, like cortical blindness.
Maybe you don't need the matter and just need the process ala XKCD Rockputers or maybe theres lower-but-real levels of consciousness more in line with panpsychic views (e.g., a neuron has a neuron's consciousness, not just being an ant in a conscious ant hive). At the very least it seems reasonable to say if you perfectly reproduced a human brain it booted it up it would have qualia.
Unless someone believes in some sort of divine spark, in evolution there is a continuum from replicating reactions-->early life-->mudskipper-->man, along which you can place a mark on where we get something "conscious". I don't think many people would be comfortable saying a dog or homo habilis isn't conscious, and they also wouldn't say an oxidation reaction or eukaryote is conscious in the way we use it. So it definitely gets messy.
A Ship of Theseus argument can go from the other direction. Instead of asking when we get something "conscious", you can ask when you lose it. Is someone with a cochlear implant less conscious/human? What if they had some treatment after a stroke that used pig neurons, or functionally identical silicon parts? How many neurons would you have to replace that way to change that? What if you completely replace the brain with those identical silicon parts?
Thanks for the link. I read through the post (twice even!), but I'm not sure that it really addresses my question. Actually, it seems to sidestep it. I think this is probably the key conclusion of...
Thanks for the link. I read through the post (twice even!), but I'm not sure that it really addresses my question. Actually, it seems to sidestep it. I think this is probably the key conclusion of the post
[If a] model is also implicitly a model of the world, [...] we shouldn't surprised that such a model grants a kind of "understanding" about how the world works.”
But this statement feels somewhat non-committal, and honestly I think most people would be surprised to learn that the xkcd rock-pushers are creating conscious beings so long as there's an isomorphism between the rocks and a conceivable universe.
To give a more concrete example (well, maybe not), in classical mechanics one can describe the dynamics of a system in position space (Lagrangian mechanics). This is a (relatively) intuitive way of describing a system: we're all familiar with what it "looks like" for a ball to fly through the air on some trajectory. However, it's also possible to re-express this system in terms of phase space, a vector space with double the dimensions of positional space but which treats momentum and position on equal footing (Hamiltonian mechanics). However, if you went to a philosopher and suggested that we don't live in coordinate space but actually phase space, they'd be somewhat dubious (speaking from personal experience). The two models might be isomorphic, but one certainly feels "more real" than the other.
But I dunno, that's just my reading of it. What about the post speaks to you?
I think the most relevant part is found at the end: At least that was why I suggested it. In hindsight it might have been a poor fit in answer to your post. More relevant if I had suggested it to...
Thanks for the link. I read through the post (twice even!), but I'm not sure that it really addresses my question. Actually, it seems to sidestep it.
I think the most relevant part is found at the end:
qualia (is) poorly understood. (...) Currently, we have no real leads in solving the hard problem. Thus who can say that there couldn’t be hypothetical language models that feel the wordiness of certain kinds of words? Maybe verbs are sharp and adjectives are so. We haven’t got a theory of qualia that would rule this out.
At least that was why I suggested it. In hindsight it might have been a poor fit in answer to your post. More relevant if I had suggested it to user RheingoldRiver.
But this statement feels somewhat non-committal, and honestly I think most people would be surprised to learn that the xkcd rock-pushers are creating conscious beings so long as there's an isomorphism between the rocks and a conceivable universe.
I personnaly think a rock simulation person would be just as real and counscious as I am. I always read the XKCD comic as implying that the author does too. The rock people can say things like "I am counscious", why would you doubt them but trust me when I say it? Regardless what any of us think, do you feel like our intuition (which are based on broad generalisation of our past experience which do not usually include rock people ) apply here?
To give a more concrete example (well, maybe not), in classical mechanics one can describe the dynamics of a system in position space (Lagrangian mechanics). This is a (relatively) intuitive way of describing a system: we're all familiar with what it "looks like" for a ball to fly through the air on some trajectory. However, it's also possible to re-express this system in terms of phase space, a vector space with double the dimensions of positional space but which treats momentum and position on equal footing (Hamiltonian mechanics). However, if you went to a philosopher and suggested that we don't live in coordinate space but actually phase space, they'd be somewhat dubious (speaking from personal experience). The two models might be isomorphic, but one certainly feels "more real" than the other.
I'm pretty sure you're baiting me here cause the answer seems obvious... I'll bite: I think one models seems more real because that's how our brain naturally interprets it. If our brain modelled the world using hamiltonian mechanics, we would find that more real. Similarly, my brain cannot think of "gravity is bending space-time itself" as real without using lots of métaphores, because it (the brain) evolved at scales were it doesn't impact me enough to be worth modeling intuitively, and yet i'm told (I'm not a physicist, it took me 90 minutes of reading about hamiltonian mechanics to realise I still don't really understand how it works) it really is the case.
PS: I've found this also somewhat relevant fiction story. It's old but short and fun. It's also from an author I really like the style of.
Full disclosure, I think I mostly agree with your views. I'm not trying to waste your time playing Devil's advocate, but for me at least, these are very much my beliefs -- I don't think my beliefs...
Full disclosure, I think I mostly agree with your views. I'm not trying to waste your time playing Devil's advocate, but for me at least, these are very much my beliefs -- I don't think my beliefs are well-justified besides being intuitive. Hence my pushback: I'm trying to fill in the gaps.
I personnaly think a rock simulation person would be just as real and counscious as I am. I always read the XKCD comic as implying that the author does too.
The next question, then, is how far do you want to push this idea? Suppose I don't actually place the rocks down; suppose instead I just write down all the rules for the simulation and what their initial conditions should be. Or suppose I don't even both to write down the initial conditions. Is that somehow less real than the rock simulation? Taken to its logical conclusion, you have Tegmark's mathematical universe hypothesis.
I'm pretty sure you're baiting me here cause the answer seems obvious... I'll bite: I think one models seems more real because that's how our brain naturally interprets it.
Perhaps I'm baiting you a little bit. :p
I think the objection here is that while both models are perfectly good descriptions of the physical world, most people actually expect the world to be something. Mathematical models do an excellent job describing the world (albeit partially because it's really the only tool we have), but just because something has a mathematical description doesn't mean it's inherently mathematical. Or as Alfred Korzybski put it, the map is not the territory.
As a less heady analogy than Hamiltonian vs Lagrangian mechanics, let's suppose you have a cardboard cutout of a triangle in front of you. We can describe that triangle in many ways, but one way of describing it is by its symmetries (or at least approximate symmetries, since the triangle is not perfectly symmetrical), e.g. its rotational symmetry.
The rotational symmetry of a triangle can be described by the cyclic group Z_3 (i.e., addition modulo 3). But if I told you that the triangle before you is actually just the group ((0, 1, 2), +%3), you could prove me wrong by hitting me your improvised shuriken. Of course, we know that this description is too simplistic -- your triangle has other symmetries (e.g., reflection symmetry) as well as all the gauge "symmetries" of the Standard Model. That is, if you want to describe your triangle exactly, the mathematical description is going to be rather complicated.
But what if I started adding other, untestable restrictions on your triangle (eg, your triangle actually lies in 26 dimensional space, but those dimensions are too tightly wrapped to observe)? I can posit different restrictions that are incompatible. Most people would expect your cardboard triangle to exist in some particular reality, not simultaneously in incompatible ones.
Returning to the question of Hamiltonian vs Lagrangian mechanics, even if both models are isomorphic (in the sense that they both make the same predictions), it does assume that fundamentally there is some sort of underlying vector space that we live in -- either N-dimensional coordinate space (Lagrangian mechanics) or 2 N-dimensional phase space in the case of Hamiltonian mechanics. It might not be possible to distinguish between the two scenarios (since the predictions are the same), but it does feel like they actually are different things.
Yes. I believe the xkcd comic shows how a universe is something from nothing, more or less. Our universe, at its most base layer, is supposed to be very similar to those rocks. Just some quantized...
Yes. I believe the xkcd comic shows how a universe is something from nothing, more or less. Our universe, at its most base layer, is supposed to be very similar to those rocks. Just some quantized fluctuations in varying fields. Sometimes the fields interact with each other.
I’ve got a pet theory about the origin of our universe. It could be that there was a universe, with laws of physics similar to ours. This universe has stuff like matter. Something like gravity. But it’s gone through a heat death. Everything is spread out, individual alt-atoms floating in space. These alt-atoms are our quarks and subatomic particles. The reason we have such different physics on a quantum level is those are the physics of the base universe. Our macro physics are all emergent properties. So when you think - you’re really just waiting a long long time for these atoms in a dead universe to hit each other in a certain pattern. They say at a small scale everything’s really spread out. Electrons are far from their nuclei. Things get hard to measure - maybe because we’re measuring emergent properties with emergent properties. Like trying to measure a modern computer’s transistors through the behavior of a NES emulator.
Consciousness is a human obsession. It’s the base layer to our perception, with things like analytical thought on top of it. We wonder how can anything mundane, like atoms, be a layer below it. I don’t need to and can’t explain how - it’s really the sum of a googol particle interactions. We already know that we’re made of matter. And if we’re more than matter the extra stuff will also be mundane.
As I asked Peppeto: The next question, then, is how far do you want to push this idea? Suppose I don't actually place the rocks down; suppose instead I just write down all the rules for the...
Yes. I believe the xkcd comic shows how a universe is something from nothing, more or less. Our universe, at its most base layer, is supposed to be very similar to those rocks.
The next question, then, is how far do you want to push this idea? Suppose I don't actually place the rocks down; suppose instead I just write down all the rules for the simulation and what their initial conditions should be. Or suppose I don't even both to write down the initial conditions. Is that somehow less real than the rock simulation? Taken to its logical conclusion, you have Tegmark's mathematical universe hypothesis.
I’ve got a pet theory about the origin of our universe. It could be that there was a universe, with laws of physics similar to ours. This universe has stuff like matter. Something like gravity. But it’s gone through a heat death. Everything is spread out, individual alt-atoms floating in space. These alt-atoms are our quarks and subatomic particles. The reason we have such different physics on a quantum level is those are the physics of the base universe. Our macro physics are all emergent properties. So when you think - you’re really just waiting a long long time for these atoms in a dead universe to hit each other in a certain pattern. They say at a small scale everything’s really spread out. Electrons are far from their nuclei. Things get hard to measure - maybe because we’re measuring emergent properties with emergent properties. Like trying to measure a modern computer’s transistors through the behavior of a NES emulator.
Yours is a fun theory, and it actually reminds me of one of my old pet ideas (I won't be so bold as to call it a theory): that shadow universes exist in dark matter. You can imagine a universe (or even multiple universes) almost exactly like ours all lying on top of each other, except instead of being composed of normal matter described by the Standard Model, they're composed of dark matter described by different quantum field theories that don't couple with ours. It just so happens that these universes can still feel each other gravitationally.
Unfortunately my idea doesn't really work out (at least, not in any interesting sort of way) since dark matter doesn't clump, suggesting that dark matter rarely interacts with itself. Oh well!
But if you want my two cents on your pet theory (consider it a bargain -- Sabine Hossenfelder charges $50 / 20 minutes for this!), I think it probably raises more questions than it answers. Still, I think I see the logic in it. Essentially you're trying to explain why the universe is fundamentally random, right? I think this solution addresses one oddity of quantum mechanics, but it doesn't address the more troubling peculiarities like Wheeler's delayed-choice experiment. I'd recommend ruminating on delayed-choice experiments for a bit and trying to square it with your mental model.
As for the mathematical universe - if properly modeled the humans would behave identically. Something evaluating the model could detect the thoughts the humans have. The humans would inevitably...
As for the mathematical universe - if properly modeled the humans would behave identically. Something evaluating the model could detect the thoughts the humans have. The humans would inevitably have a conversation about the nature of the universe and the definition of consciousness. The humans would necessarily perceive things, think, etc.
Just to be clear, the mathematical universe hypothesis doesn't merely posit that the universe has a mathematical description -- it posits that the universe actually is a mathematical object and...
Just to be clear, the mathematical universe hypothesis doesn't merely posit that the universe has a mathematical description -- it posits that the universe actually is a mathematical object and that all mathematical objects are "real".
For instance, since there exists a mathematical description of triangles on a plane stretching, moving, and rotating over time, this would imply the existence of a universe of triangles doing... triangle things. Moreover, there exists a more general mathematical structure describing all possible configurations of triangles on a plane evolving over time, and since this is also a mathematical structure, the mathematical universe hypothesis would suggest that this generalized universe also exists (as bizarre as that would "look").
So here's an idea that kinda fits with yours. Every human has created an ego tunnel that is based on our consciousness combining all of the senses our bodies possess into a model of reality in our...
So here's an idea that kinda fits with yours. Every human has created an ego tunnel that is based on our consciousness combining all of the senses our bodies possess into a model of reality in our minds. There very well could be and is much more out there that we don't have in our mental model because our senses don't process anything beyond their scope. We know that there is more out there because humans have created sensors that detect things our biological senses cannot detect, but who is to say that these imperceptible parts of the universe are not very detectable by something that is outside the realm of our biological senses or our built sensors.
We could currently be surrounded by beings that exist in our universe but are made of something that we can't sense or interact with. Likewise, they cannot sense or interact with what our world model is made of. Their conscious world model doesn't include us but occupies the same space.
Your idea reminds me a bit of Anathem. I've been wondering something similar recently: if consciousness is just a process, then are two consciousnesses indistinguishable if the processes are...
Your idea reminds me a bit of Anathem.
I've been wondering something similar recently: if consciousness is just a process, then are two consciousnesses indistinguishable if the processes are identical? For instance, if we had a copy of our universe that was identical within the observable universe but differed outside it, would the "same" conscious being in both universes have the same consciousness? Would they share the same consciousness?
What if only imperceptible differences exist (e.g., identical universes except a molecule of helium has been displaced by 0.1 fm somewhere in the sun)? If two otherwise identical consciousnesses can't observe the difference, are the two consciousnesses indistinguishable?
This statement can only possibly be coherent if you accompany it with a rigorous definition of "think" which doesn't circularly refer back to human brains. The dictionary definition is "to...
This statement can only possibly be coherent if you accompany it with a rigorous definition of "think" which doesn't circularly refer back to human brains.
The dictionary definition is "to exercise the the powers of of judgement, conception, or inference", which computers without question can be programmed to do. (Rigorous logical reasoning is something we were programming computers to do way back at the very beginning of AI research, in the '50s and '60s.)
Part of the reason any discussion of AI is super hard to have is that almost literally none of the terms or concepts involved are well-defined. What even is "intelligence", literally half the initialism! Nobody's come up with a rigorous definition yet, so if you have one, you stand to revolutionize the field if you publish it.
I found this article that was refernced in the article posted by OP https://newatlas.com/science/dishbrain-cells-sentience-play-pong/?itm_source=newatlas&itm_medium=article-body With the following...
I found this article that was refernced in the article posted by OP
“Don’t worry, while these dishes of neurons can change their responses based on stimulation, they are not sci-fi style intelligence in a dish, these are simple (albeit interesting and scientifically important) circuit responses,” said Professor Tara Spires-Jones, UK Dementia Research Institute Programme Lead.
So I don't think you have to worry quite yet.
Some more specific details:
Training DishBrain was understandably tricky. Reward and punishment are usually key if you’re going to train an organism to perform a specific task, but that won’t work on these cell cultures because they have no dopamine system to incentivize them. Instead, the researchers took advantage of what’s called the free energy principle.
Essentially, this idea says that cells at this level will always try to minimize the unpredictability in their environment. So, if the brain cells failed to hit the ball with their paddle, the system would deliver an unpredictable stimulus for four seconds, but for every successful hit they would receive a brief, predictable signal before the game continued on, also in a predictable manner. Using this, DishBrain learned to play the game within five minutes.
“The beautiful and pioneering aspect of this work rests on equipping the neurons with sensations – the feedback – and crucially the ability to act on their world,” said Professor Karl Friston, co-author of the study. “Remarkably, the cultures learned how to make their world more predictable by acting upon it. This is remarkable because you cannot teach this kind of self-organization; simply because – unlike a pet – these mini brains have no sense of reward and punishment.”
I wish I had an answer to your question. The article says, but I don't know enough to know how that differs from traditional chips, or if the claim in correct.
I wish I had an answer to your question. The article says,
The technology promises machines that can continue to learn new abilities without compromising old ones, that can adapt well to change, and that can map old knowledge onto new situations – while continually self-optimizing their use of computing power, memory and energy.
but I don't know enough to know how that differs from traditional chips, or if the claim in correct.
All I can think as I read this was the beginning of true androids!! Of course, that's definitely not it and the brain-comouter interface stuff is the most exciting in that it could have wondrous...
All I can think as I read this was the beginning of true androids!!
Of course, that's definitely not it and the brain-comouter interface stuff is the most exciting in that it could have wondrous medical uses!
I scanned the preprint paper. Looks like they aren't using human and mouse cells in the same system. They've build different systems using mouse embryo brain cells OR human pluripotent stem cells....
I scanned the preprint paper. Looks like they aren't using human and mouse cells in the same system. They've build different systems using mouse embryo brain cells OR human pluripotent stem cells.
I looked over the paper only very quickly, but looks like part of their experiment was to compare the activity and function between the different cell culture types.
In sum, I think it'd be fair to say they were figuring out what they needed :)
This is just speculation, but mouse brain tissue is much cheaper and (potentially) easier to work with than that of humans. Maybe it's for experiments and prototyping?
This is just speculation, but mouse brain tissue is much cheaper and (potentially) easier to work with than that of humans. Maybe it's for experiments and prototyping?
I for one cannot wait for this stuff, I always feel like I was born a few thousand years to early. Integration into computers and then eventually defying death being the end goal of all of this...
I for one cannot wait for this stuff, I always feel like I was born a few thousand years to early. Integration into computers and then eventually defying death being the end goal of all of this would be amazing.
I'm currently reading Martha Wells Murderbot books. Fun but pulpy sci-fi. But just being a cyborg doesn't seem to eliminate trouble from life in Wells' model.
I'm currently reading Martha Wells Murderbot books. Fun but pulpy sci-fi. But just being a cyborg doesn't seem to eliminate trouble from life in Wells' model.
Interesting, but one big problem I can see is that traditional computational materials have an entirely different set of needs in comparison to living tissue, often contradictory needs. (Plenty of...
Interesting, but one big problem I can see is that traditional computational materials have an entirely different set of needs in comparison to living tissue, often contradictory needs. (Plenty of ethical questions too, but I'm going to limit it to the practical for sake of brevity)
All the contemporary computer needs is an electrical source and good airflow. Brain tissue is going to need food, sleep, waste removal, a stricter operating temperature regime, and likely many more things I can't begin to fathom.
I am both excited and scared by this. I don't know why. I think maybe I've read too much dystopian fiction.
What are the implications of this? What are the possibilities for this technology?
The big thing I can think of is that silicone, by itself, can't think it can only process. We (humans) have come up with clever tricks to emulate thinking, but (assuming this does what's claimed) this would truly allow computers to think like humans. The implications of that are pretty far-reaching. Think self-driving vehicles or robotics. The SUPER interesting bit to me is this combination of traditional silicone with brain tissue here. It makes me wonder if we can have the best of both worlds with an an AI that can truly think with a computer that can logically make the best decision. Not an expert in any related field (unless you count IT) so those are more my musing than anything else, lol.
Silicon, not silicone. Not that that's super important - just a common typo :D
I don't think it's fair to say silicon can only process. What if we had a computer powerful enough to simulate a large amount of matter? And then have the computer simulate a full human body and brain. Would it not be thinking?
Or maybe we have a computer simulate all of the neurons in a brain. Is it thinking?
Or what if our non-biologically designed software neural networks get powerful enough to do all of the things a human mind can. Is it thinking?
No, it wouldn't. The analogy I like is this: Imagine you have 2 people who have learned to give the answers to multiplication problems. Alice learned to do column multiplication. Bob has memorized a multiplication table all the way out to 100,000 x 100,000. Most of the time, it'll be impossible to detect the difference between Alice and Bob solving (though Bob might be a lot faster). But Bob is not really multiplying, whereas Alice is. And to further the analogy, we can make Bob's simulation of multiplying even better, by teaching him all the way to 1,000,000 (Bob has a very good memory). Yet still only Alice is capable of true multiplication, able to multiply any numbers we give her. But Bob is an excellent approximation, impossible to tell the difference most of the time.
I don't think that's a good analogy. In teaearlgraycold's hypothetical, the simulated brain would be doing exactly the same things the biological brain is doing.
I see no difference between a real wet human brain and a perfectly simulated silicon one. As far as we know, consciousness is an emergent property of the complex systems in our brains, and if a computer could perfectly simulate those systems, then there's no reason to say that the consciousness that emerges is any less legitimate. The only objection to that would be that humans can think because we have souls, which as a naturalist, I don't believe.
Ok, that's fair. I guess I'm approaching from the POV that this hypothetical seems incredibly unrealistic, because every attempt we've made to model so far doesn't look anything like what humans do, and to a large extent we don't even know what humans do enough to be able to intentionally model it (to the best of my knowledge at least). In fact I consider it somewhat of a paradox if we were able to intentionally model true thinking, because that means there's a 0% chance that we're not intentionally modeled by a higher-order universe, with infinitely many higher-order universes above.
I feel like I'm getting into the weeds here, but the fact that we can't currently simulate a brain doesn't mean it isn't possible; it only means that we can't do it yet. While the inclusion of brain cells into electronics could speed up the development of truly "thinking" machines, I was simply agreeing with and defending teaearlgraycold's statement that it's a bit unfair to say that a computer made with brain cells can do things that silicon fundamentally cannot.
I don't see how this follows. That we can intentionally model true thinking doesn't say anything about our own development as biological beings. We were still shaped by billions of years of evolution, and the development of intelligent synthetic life wouldn't change that.
I would like to throw at this discussion the fact that we don't understand the brain 100%, and we can't model something if we don't understand how it works. We can't even tell right now if it's possible with the current limitations in Computer Science, because we don't understand what we want to simulate, so we don't know what we need to simulate it correctly.
So, it could be possible, but there's also the possibility that we can't with our technology, and to be able to really do it, before we build an even better computer we have to understand how we think.
One example is about how language makes us able to think, what's the relationship between language, culture and individual thinking, what would happen to a human that didn't learn any language whatsoever, and so on.
We can't copy from scratch what we can't understand.
It would be more simulation than modeling. Modeling would mean an approximation. We’d probably model the cells but copy the connections from a brain scan.
I think that’s a good argument for why a simulated human is not a real human. But not why a machine can not think - unless we define thinking as exclusively what a real human brain does.
If a simulation is functionally equivalent to "the real thing," then what's the difference? It doesn't even really make sense to make a philosophical distinction, as philosophy divorced from external reality is just navel-gazing debate about angels boogieing down on pinheads.
We don't know exactly what gives rise to sapience, though I think it likely to be an emergent phenomenon of complex intra-acting systems. If that's true, the difference between a human brain and a silicon semiconductor-based thinking machine is only a matter of degree, not kind.
I do wonder how far you can push this analogy. Take the xkcd "A Bunch of Rocks". If you could create a simulation of the universe just using a limitless supply of rocks, does that mean you could simulate consciousness too? If so, where would the consciousness reside, so to speak?
Personally in the camp of materialism being a lower bounds. Matter structured and changing in a particular way appears to give rise to consciousness. Neurologists could see a brain scan of someone with some particularly well-documented damage and infer something about their qualia, like cortical blindness.
Maybe you don't need the matter and just need the process ala XKCD Rockputers or maybe theres lower-but-real levels of consciousness more in line with panpsychic views (e.g., a neuron has a neuron's consciousness, not just being an ant in a conscious ant hive). At the very least it seems reasonable to say if you perfectly reproduced a human brain it booted it up it would have qualia.
Unless someone believes in some sort of divine spark, in evolution there is a continuum from replicating reactions-->early life-->mudskipper-->man, along which you can place a mark on where we get something "conscious". I don't think many people would be comfortable saying a dog or homo habilis isn't conscious, and they also wouldn't say an oxidation reaction or eukaryote is conscious in the way we use it. So it definitely gets messy.
A Ship of Theseus argument can go from the other direction. Instead of asking when we get something "conscious", you can ask when you lose it. Is someone with a cochlear implant less conscious/human? What if they had some treatment after a stroke that used pig neurons, or functionally identical silicon parts? How many neurons would you have to replace that way to change that? What if you completely replace the brain with those identical silicon parts?
You might find this interresting.
(I am not the author.)
Thanks for the link. I read through the post (twice even!), but I'm not sure that it really addresses my question. Actually, it seems to sidestep it. I think this is probably the key conclusion of the post
But this statement feels somewhat non-committal, and honestly I think most people would be surprised to learn that the xkcd rock-pushers are creating conscious beings so long as there's an isomorphism between the rocks and a conceivable universe.
To give a more concrete example (well, maybe not), in classical mechanics one can describe the dynamics of a system in position space (Lagrangian mechanics). This is a (relatively) intuitive way of describing a system: we're all familiar with what it "looks like" for a ball to fly through the air on some trajectory. However, it's also possible to re-express this system in terms of phase space, a vector space with double the dimensions of positional space but which treats momentum and position on equal footing (Hamiltonian mechanics). However, if you went to a philosopher and suggested that we don't live in coordinate space but actually phase space, they'd be somewhat dubious (speaking from personal experience). The two models might be isomorphic, but one certainly feels "more real" than the other.
But I dunno, that's just my reading of it. What about the post speaks to you?
I think the most relevant part is found at the end:
At least that was why I suggested it. In hindsight it might have been a poor fit in answer to your post. More relevant if I had suggested it to user RheingoldRiver.
I personnaly think a rock simulation person would be just as real and counscious as I am. I always read the XKCD comic as implying that the author does too. The rock people can say things like "I am counscious", why would you doubt them but trust me when I say it? Regardless what any of us think, do you feel like our intuition (which are based on broad generalisation of our past experience which do not usually include rock people ) apply here?
I'm pretty sure you're baiting me here cause the answer seems obvious... I'll bite: I think one models seems more real because that's how our brain naturally interprets it. If our brain modelled the world using hamiltonian mechanics, we would find that more real. Similarly, my brain cannot think of "gravity is bending space-time itself" as real without using lots of métaphores, because it (the brain) evolved at scales were it doesn't impact me enough to be worth modeling intuitively, and yet i'm told (I'm not a physicist, it took me 90 minutes of reading about hamiltonian mechanics to realise I still don't really understand how it works) it really is the case.
PS: I've found this also somewhat relevant fiction story. It's old but short and fun. It's also from an author I really like the style of.
Full disclosure, I think I mostly agree with your views. I'm not trying to waste your time playing Devil's advocate, but for me at least, these are very much my beliefs -- I don't think my beliefs are well-justified besides being intuitive. Hence my pushback: I'm trying to fill in the gaps.
The next question, then, is how far do you want to push this idea? Suppose I don't actually place the rocks down; suppose instead I just write down all the rules for the simulation and what their initial conditions should be. Or suppose I don't even both to write down the initial conditions. Is that somehow less real than the rock simulation? Taken to its logical conclusion, you have Tegmark's mathematical universe hypothesis.
Perhaps I'm baiting you a little bit. :p
I think the objection here is that while both models are perfectly good descriptions of the physical world, most people actually expect the world to be something. Mathematical models do an excellent job describing the world (albeit partially because it's really the only tool we have), but just because something has a mathematical description doesn't mean it's inherently mathematical. Or as Alfred Korzybski put it, the map is not the territory.
As a less heady analogy than Hamiltonian vs Lagrangian mechanics, let's suppose you have a cardboard cutout of a triangle in front of you. We can describe that triangle in many ways, but one way of describing it is by its symmetries (or at least approximate symmetries, since the triangle is not perfectly symmetrical), e.g. its rotational symmetry.
The rotational symmetry of a triangle can be described by the cyclic group Z_3 (i.e., addition modulo 3). But if I told you that the triangle before you is actually just the group ((0, 1, 2), +%3), you could prove me wrong by hitting me your improvised shuriken. Of course, we know that this description is too simplistic -- your triangle has other symmetries (e.g., reflection symmetry) as well as all the gauge "symmetries" of the Standard Model. That is, if you want to describe your triangle exactly, the mathematical description is going to be rather complicated.
But what if I started adding other, untestable restrictions on your triangle (eg, your triangle actually lies in 26 dimensional space, but those dimensions are too tightly wrapped to observe)? I can posit different restrictions that are incompatible. Most people would expect your cardboard triangle to exist in some particular reality, not simultaneously in incompatible ones.
Returning to the question of Hamiltonian vs Lagrangian mechanics, even if both models are isomorphic (in the sense that they both make the same predictions), it does assume that fundamentally there is some sort of underlying vector space that we live in -- either N-dimensional coordinate space (Lagrangian mechanics) or 2 N-dimensional phase space in the case of Hamiltonian mechanics. It might not be possible to distinguish between the two scenarios (since the predictions are the same), but it does feel like they actually are different things.
Yes. I believe the xkcd comic shows how a universe is something from nothing, more or less. Our universe, at its most base layer, is supposed to be very similar to those rocks. Just some quantized fluctuations in varying fields. Sometimes the fields interact with each other.
I’ve got a pet theory about the origin of our universe. It could be that there was a universe, with laws of physics similar to ours. This universe has stuff like matter. Something like gravity. But it’s gone through a heat death. Everything is spread out, individual alt-atoms floating in space. These alt-atoms are our quarks and subatomic particles. The reason we have such different physics on a quantum level is those are the physics of the base universe. Our macro physics are all emergent properties. So when you think - you’re really just waiting a long long time for these atoms in a dead universe to hit each other in a certain pattern. They say at a small scale everything’s really spread out. Electrons are far from their nuclei. Things get hard to measure - maybe because we’re measuring emergent properties with emergent properties. Like trying to measure a modern computer’s transistors through the behavior of a NES emulator.
Consciousness is a human obsession. It’s the base layer to our perception, with things like analytical thought on top of it. We wonder how can anything mundane, like atoms, be a layer below it. I don’t need to and can’t explain how - it’s really the sum of a googol particle interactions. We already know that we’re made of matter. And if we’re more than matter the extra stuff will also be mundane.
As I asked Peppeto:
The next question, then, is how far do you want to push this idea? Suppose I don't actually place the rocks down; suppose instead I just write down all the rules for the simulation and what their initial conditions should be. Or suppose I don't even both to write down the initial conditions. Is that somehow less real than the rock simulation? Taken to its logical conclusion, you have Tegmark's mathematical universe hypothesis.
Yours is a fun theory, and it actually reminds me of one of my old pet ideas (I won't be so bold as to call it a theory): that shadow universes exist in dark matter. You can imagine a universe (or even multiple universes) almost exactly like ours all lying on top of each other, except instead of being composed of normal matter described by the Standard Model, they're composed of dark matter described by different quantum field theories that don't couple with ours. It just so happens that these universes can still feel each other gravitationally.
Unfortunately my idea doesn't really work out (at least, not in any interesting sort of way) since dark matter doesn't clump, suggesting that dark matter rarely interacts with itself. Oh well!
But if you want my two cents on your pet theory (consider it a bargain -- Sabine Hossenfelder charges $50 / 20 minutes for this!), I think it probably raises more questions than it answers. Still, I think I see the logic in it. Essentially you're trying to explain why the universe is fundamentally random, right? I think this solution addresses one oddity of quantum mechanics, but it doesn't address the more troubling peculiarities like Wheeler's delayed-choice experiment. I'd recommend ruminating on delayed-choice experiments for a bit and trying to square it with your mental model.
As for the mathematical universe - if properly modeled the humans would behave identically. Something evaluating the model could detect the thoughts the humans have. The humans would inevitably have a conversation about the nature of the universe and the definition of consciousness. The humans would necessarily perceive things, think, etc.
Just to be clear, the mathematical universe hypothesis doesn't merely posit that the universe has a mathematical description -- it posits that the universe actually is a mathematical object and that all mathematical objects are "real".
For instance, since there exists a mathematical description of triangles on a plane stretching, moving, and rotating over time, this would imply the existence of a universe of triangles doing... triangle things. Moreover, there exists a more general mathematical structure describing all possible configurations of triangles on a plane evolving over time, and since this is also a mathematical structure, the mathematical universe hypothesis would suggest that this generalized universe also exists (as bizarre as that would "look").
So here's an idea that kinda fits with yours. Every human has created an ego tunnel that is based on our consciousness combining all of the senses our bodies possess into a model of reality in our minds. There very well could be and is much more out there that we don't have in our mental model because our senses don't process anything beyond their scope. We know that there is more out there because humans have created sensors that detect things our biological senses cannot detect, but who is to say that these imperceptible parts of the universe are not very detectable by something that is outside the realm of our biological senses or our built sensors.
We could currently be surrounded by beings that exist in our universe but are made of something that we can't sense or interact with. Likewise, they cannot sense or interact with what our world model is made of. Their conscious world model doesn't include us but occupies the same space.
Your idea reminds me a bit of Anathem.
I've been wondering something similar recently: if consciousness is just a process, then are two consciousnesses indistinguishable if the processes are identical? For instance, if we had a copy of our universe that was identical within the observable universe but differed outside it, would the "same" conscious being in both universes have the same consciousness? Would they share the same consciousness?
What if only imperceptible differences exist (e.g., identical universes except a molecule of helium has been displaced by 0.1 fm somewhere in the sun)? If two otherwise identical consciousnesses can't observe the difference, are the two consciousnesses indistinguishable?
This statement can only possibly be coherent if you accompany it with a rigorous definition of "think" which doesn't circularly refer back to human brains.
The dictionary definition is "to exercise the the powers of of judgement, conception, or inference", which computers without question can be programmed to do. (Rigorous logical reasoning is something we were programming computers to do way back at the very beginning of AI research, in the '50s and '60s.)
Part of the reason any discussion of AI is super hard to have is that almost literally none of the terms or concepts involved are well-defined. What even is "intelligence", literally half the initialism! Nobody's come up with a rigorous definition yet, so if you have one, you stand to revolutionize the field if you publish it.
I found this article that was refernced in the article posted by OP
https://newatlas.com/science/dishbrain-cells-sentience-play-pong/?itm_source=newatlas&itm_medium=article-body
With the following quote:
So I don't think you have to worry quite yet.
Some more specific details:
I wish I had an answer to your question. The article says,
but I don't know enough to know how that differs from traditional chips, or if the claim in correct.
All I can think as I read this was the beginning of true androids!!
Of course, that's definitely not it and the brain-comouter interface stuff is the most exciting in that it could have wondrous medical uses!
What a time to be alive.
I'd like to know why they needed human and mouse brain tissue. And a lot of other things.
I scanned the preprint paper. Looks like they aren't using human and mouse cells in the same system. They've build different systems using mouse embryo brain cells OR human pluripotent stem cells.
I looked over the paper only very quickly, but looks like part of their experiment was to compare the activity and function between the different cell culture types.
In sum, I think it'd be fair to say they were figuring out what they needed :)
https://www.biorxiv.org/content/10.1101/2021.12.02.471005v2
Thank you! That makes sense. My scifi drenched brain went much weirder.
Thank you!
This is just speculation, but mouse brain tissue is much cheaper and (potentially) easier to work with than that of humans. Maybe it's for experiments and prototyping?
That's an interesting question. After your reply, I tried a quick search to find the answer but came up with nothing. :(
I for one cannot wait for this stuff, I always feel like I was born a few thousand years to early. Integration into computers and then eventually defying death being the end goal of all of this would be amazing.
I'm currently reading Martha Wells Murderbot books. Fun but pulpy sci-fi. But just being a cyborg doesn't seem to eliminate trouble from life in Wells' model.
Link to the Monash University news release
Interesting, but one big problem I can see is that traditional computational materials have an entirely different set of needs in comparison to living tissue, often contradictory needs. (Plenty of ethical questions too, but I'm going to limit it to the practical for sake of brevity)
All the contemporary computer needs is an electrical source and good airflow. Brain tissue is going to need food, sleep, waste removal, a stricter operating temperature regime, and likely many more things I can't begin to fathom.
This headline brings me back to Passages in the Void, especially the beginning of Mortal Passage.
That was a great read, thanks for linking to it!