So, basically your argument is that in simulating a brain you will by necessity imbue that simulation with real consciousness. Yeah?
If this is your argument, I want to let you know that this fruit fly is exactly as conscious as ChatGPT or Gemini. The way computational neuroscience works is that you take the wiring diagram of a brain (called a connectome) and use that map to connect a large number of artificial neurons (as in Neural Network, these are the exact same things current LLMs use).
In order to believe this fly is sapient, sentient, conscious, or even alive at all, you would need to believe that not only could LLMs become sentient, they already are. Do you believe that GPT-6 is a real, sentient person like us?
We’re talking about simulating neurons here. Replicating as closely as possible the structure and function of a brain and the components that make it up, not throwing everything into a statistics machine that spits out responses based on probability.
I understand perfectly that an LLM is not simulating a brain or any sort of thought process.
I think you maybe need to do more research on what an artificial neuron actually is, because throwing everything into a statistics machine that spits out responses based on probability is exactly what they’re doing with this Drosophila connectome. It’s just a statistics machine shaped like a fruit fly ganglion.
Yeah, I’m certainly an idiot when it comes to these things, but my thoughts lead this way:
Do you think these simulations are or would ever be good enough that you could wire them up to a fly’s body and the fly would be able to essentially be a fly, albeit with an artificial brain? What makes a meat ganglion any different to a silicone one? Where does the consciousness come from and why does the artificial brain not have the divine spark, so to speak?
I’ve said a few times that I’ve probably misunderstood the fidelity of the simulation in this article, but the question I’m posing still stands I believe.
Well, if Fleshlight® is to be believed, there’s no difference at all between meat and silicone!
But in all seriousness, no matter how good this simulation gets, you couldn’t really hook it up to an actual fly, and if you somehow found a way to do so, it would almost certainly just do either nothing or instantly kill the fly body. :( The way all computational neuroscience sims work is by connecting the wiring diagram of a brain up to a thing that AI bros call a “neuron” but is actually bunch of fancy math and transfer functions.
The math here kind of sort of resembles a biological neuron, just about as much as this emoji of a drawing of a bird resembles a real life airplane. You can imagine how they both have wings and could probably fly, but one of them is uhhhhh definitely not an airplane, it’s an emoji of a bird so why the fuck do researchers keep calling that emoji of a bird an airplane, and saying that their cool new bird emoji is literally an F-35? It’s just marketing hype. This fly simulation is no different. It’s just a bunch of bird emojis that share math outputs to other bird emojis. They don’t actually reflect anything happening in a real brain, the same way that an LLM doesn’t reflect thought.
What they’ve done here is basically putting Gemini in a 1:1 scale 3d model of a fruit fly. The impressive thing is that it can run at or close to real time on common hardware, where just a few years ago this was thoroughly in the realm of “1-2 neurons firing per minute” on the top research supercomputers on the planet.
There’s an entirely unrelated and far more upsetting field called Biocomputing, which actually does use real neurons to do computational work, and these same freaks are the ones who like sticking probes into bugs and trying to drive their bodies with electrical signals.
I mean, I think Data from TNG is a real conscious artificial person (or, you know, a TV show character played by a human actor that represents a real conscious artificial person) and I believe it’s possible that such a person could one day exist. However, the issue here is that the “neurons” they’re using for this just are not something that I or anyone else skeptical of the claims made by AI companies’ marketing departments believe could possess that divine spark that uniquely exists in conscious beings.
As an aside, I hate to admit it as a vegan but I genuinely don’t think that Drosophila spp. are what we would consider conscious. Obviously they are alive and deserving of respect, but that doesn’t really mean that everything alive is conscious. There are quite a number of very simple insects and things like roundworms that simply don’t have enough neurons to be more than biological automata that respond to basic stimuli. For example C. elegans has a mere 302 neurons in its body and is no more complex in its actions or behaviors than single celled organisms which I think we all agree aren’t conscious at all.
I guess what I’m saying is that consciousness is like pornography: I know it when I see it. I think consciousness is a gestalt and it’s definitely not stored in the connectome.
Thanks for taking the time to write that all out. I hope you understand that my concerns are less about this simulated fly and more about the what ifs posed by the possibility of artificial minds. I too believe that consciousness is probably some emergent property of sufficiently complex systems, but all to often I see people arguing that no artificial mind could ever be conscious, and that sort of absolutism gives me the ick.
Again, I’m not to concerned about this fly simulation, I just wonder what will happen if and when we have the capability to spin up actual artificial brains, and what sort of excuses are going to be thrown around to excuse the horrible things we do to them.
I think it’s going to be a long time before we get there. I’m quite convinced that this current tech is a dead end with regards to creating artificial consciousness.
So, basically your argument is that in simulating a brain you will by necessity imbue that simulation with real consciousness. Yeah?
If this is your argument, I want to let you know that this fruit fly is exactly as conscious as ChatGPT or Gemini. The way computational neuroscience works is that you take the wiring diagram of a brain (called a connectome) and use that map to connect a large number of artificial neurons (as in Neural Network, these are the exact same things current LLMs use).
In order to believe this fly is sapient, sentient, conscious, or even alive at all, you would need to believe that not only could LLMs become sentient, they already are. Do you believe that GPT-6 is a real, sentient person like us?
No because an llm isn’t simulating a brain.
We’re talking about simulating neurons here. Replicating as closely as possible the structure and function of a brain and the components that make it up, not throwing everything into a statistics machine that spits out responses based on probability.
Do you understand the difference?
I understand perfectly that an LLM is not simulating a brain or any sort of thought process.
I think you maybe need to do more research on what an artificial neuron actually is, because throwing everything into a statistics machine that spits out responses based on probability is exactly what they’re doing with this Drosophila connectome. It’s just a statistics machine shaped like a fruit fly ganglion.
Yeah, I’m certainly an idiot when it comes to these things, but my thoughts lead this way:
Do you think these simulations are or would ever be good enough that you could wire them up to a fly’s body and the fly would be able to essentially be a fly, albeit with an artificial brain? What makes a meat ganglion any different to a silicone one? Where does the consciousness come from and why does the artificial brain not have the divine spark, so to speak?
I’ve said a few times that I’ve probably misunderstood the fidelity of the simulation in this article, but the question I’m posing still stands I believe.
Well, if Fleshlight® is to be believed, there’s no difference at all between meat and silicone!
But in all seriousness, no matter how good this simulation gets, you couldn’t really hook it up to an actual fly, and if you somehow found a way to do so, it would almost certainly just do either nothing or instantly kill the fly body. :( The way all computational neuroscience sims work is by connecting the wiring diagram of a brain up to a thing that AI bros call a “neuron” but is actually bunch of fancy math and transfer functions.
The math here kind of sort of resembles a biological neuron, just about as much as this emoji of a drawing of a bird
resembles a real life airplane. You can imagine how they both have wings and could probably fly, but one of them is uhhhhh definitely not an airplane, it’s an emoji of a bird so why the fuck do researchers keep calling that emoji of a bird an airplane, and saying that their cool new bird emoji is literally an F-35? It’s just marketing hype. This fly simulation is no different. It’s just a bunch of bird emojis that share math outputs to other bird emojis. They don’t actually reflect anything happening in a real brain, the same way that an LLM doesn’t reflect thought.
What they’ve done here is basically putting Gemini in a 1:1 scale 3d model of a fruit fly. The impressive thing is that it can run at or close to real time on common hardware, where just a few years ago this was thoroughly in the realm of “1-2 neurons firing per minute” on the top research supercomputers on the planet.
There’s an entirely unrelated and far more upsetting field called Biocomputing, which actually does use real neurons to do computational work, and these same freaks are the ones who like sticking probes into bugs and trying to drive their bodies with electrical signals.
I mean, I think Data from TNG is a real conscious artificial person (or, you know, a TV show character played by a human actor that represents a real conscious artificial person) and I believe it’s possible that such a person could one day exist. However, the issue here is that the “neurons” they’re using for this just are not something that I or anyone else skeptical of the claims made by AI companies’ marketing departments believe could possess that divine spark that uniquely exists in conscious beings.
As an aside, I hate to admit it as a vegan but I genuinely don’t think that Drosophila spp. are what we would consider conscious. Obviously they are alive and deserving of respect, but that doesn’t really mean that everything alive is conscious. There are quite a number of very simple insects and things like roundworms that simply don’t have enough neurons to be more than biological automata that respond to basic stimuli. For example C. elegans has a mere 302 neurons in its body and is no more complex in its actions or behaviors than single celled organisms which I think we all agree aren’t conscious at all.
I guess what I’m saying is that consciousness is like pornography: I know it when I see it. I think consciousness is a gestalt and it’s definitely not stored in the connectome.
Are you saying I can hook a fruit fly brain to my fleshlight?
Thanks for taking the time to write that all out. I hope you understand that my concerns are less about this simulated fly and more about the what ifs posed by the possibility of artificial minds. I too believe that consciousness is probably some emergent property of sufficiently complex systems, but all to often I see people arguing that no artificial mind could ever be conscious, and that sort of absolutism gives me the ick.
Again, I’m not to concerned about this fly simulation, I just wonder what will happen if and when we have the capability to spin up actual artificial brains, and what sort of excuses are going to be thrown around to excuse the horrible things we do to them.
I think it’s going to be a long time before we get there. I’m quite convinced that this current tech is a dead end with regards to creating artificial consciousness.