> Our main goal is to build the world's first virtual organism - an in-silico implementation of a living creature - for the purpose of achieving an understanding of the events and mechanisms of living cells.
This blog post is interesting: "Whole Brain Emulation: No Progress on C. elegans After 10 Years" https://www.lesswrong.com/posts/mHqQxwKuzZS69CXX5/whole-brai... As other mentioned in the comments of the post, it is certainly also a matter of funding, but still I think it is very interesting how we still struggle to simulate a 302 cell worm, while some people expect an artificial superintelligence in the next ten years.
Replicating something exactly is a lot harder than making something that does the same thing in a different way. We can build planes that travel far faster than a bird could but that doesn’t mean it’s easy for us to exactly replicate how a bird flies, so I don’t think this says anything about how far away we are from a superintelligence
It's not only about exact simulation. And we actually do understand how birds can fly on a mechanical basis. This is not true for C.elegans' inner workings.
This. Submarines don't swim like fish, airplanes don't fly like birds and cars don't run like gepards.
Edit: I am surprised at the downvotes. In general, we learn from the nature, but aping it usually proved too difficult and often unpractical at the same time. Do we really want to replicate worm intelligence for practical purposes, or do we want something different?
I would say that a machine which can, say, analyze chemical compounds for their potential biological functions, is a very practical form of "intelligence" and yet very far from any biological intelligence that was ever produced in vivo. Worms cannot do that and even humans struggle with such tasks.
Yeah, but we understand why birds can fly and why fish can swim and how gepards can run, at least on a mechanical basis.
This is not true for how this 302 cell organism works. We don't know and struggle to understand. That's actually the reason why the project exists. To find out how everything works with an bottom-up approach.
While we may find shortcuts or even superior forms of intelligence without understanding how intelligence works in biological creatures, it is still curious how we struggle even with a "simple" organism like C.elegans.
No but we observably do know that that the connections between cells are important, to the point that by mimicking that we've derived significant benefit.
The Wright flyer didn't flap, and the wings only superficially look like anything a bird has.
> No but we observably do know that that the connections between cells are important, to the point that by mimicking that we've derived significant benefit.
That is true and it shows even more how important observabilty is for science and engineering. That's also why a simulation that actually provides an accurate enough model of reality might help us so much. The problem with AI right now is, that we try or even claim to understand Unix by mimicking the functionality of transistors.
> The Wright flyer didn't flap, and the wings only superficially look like anything a bird has.
They tried to mimick bird wings when coming up with flight control mechanisms.
Not only, it's also because of the square/cube law making it impossible to have a 50 ton flyer that flies in the same manner as a 5 kg flyer, you can't simply scale things up.
In defense of the project: they’ve been working a lot over the last 10 years. They recently published what they call a “liquid neural network” where they’ve isolated the neuron structure doing the work. It outperforms CNNs for some niche tasks, and it uses a fraction of the resources.
Just to make this clear, I think this project is actually one of the few sane engineering approaches for starting to understand how our brain works. Sure, claims of simulating a cat brain or even a human brain in some manner are much more catchy, but it seems to be pretty arrogant if we struggle even with C.elegans.
My cellular biochemistry does a pretty good job of replicating itself with absolutely no understanding whatsoever of how to resolve the contradictions between general relativity and quantum mechanics.
When it comes to brains, I don't know if anyone knows what might be the simplest sufficient model that would usefully replicate them, even if you specify "usefully" well enough to know if this is about fundamentals of intelligence or about the impact of drugs on cognition, which are two completely different standards.
For example, perceptrons are a toy model, but modern AI can do more in (breadth XOR single-skill performance in various domains) than any single human, even with much smaller parameter counts than we have synapses; but the broad-skilled ones also mess up in inhuman ways, like being equally good at advanced calculus as basic arithmetic, or being a poet at the level of stereotypical teenager but in every language simultaneously.
If anyone's made a neutral network that can get high on simulated caffeine — and I'm not saying it hasn't been done — it's not reached any of the places I follow discussions on this kind of thing. (Google didn't help, results were about software named Caffeine and non-artificial neurones).
I'm not saying it isn't physics, I'm saying it doesn't know what the physics is.
We're physics too, but we don't know how it all fits together.
If a sub-part of us that knows less than we do can make a copy of us, despite not knowing how it all works, that's an existence proof that we don't need to understand how it all works to make a copy of us.
If the "something" never exploits a particular law, then in principle you could simulate it if you knew all the laws that actually applied (imagine a hypothetical biological organism that never exploited or experienced quantum entanglement).
But strictly speaking, as we understand it, it's not possible to replicate something exactly without recapitulating the exact laws and running a deterministic simulation, which is not practical.
I don't think anybody is really attempting to exactly replicate things, but rather to create a physical model which can be calcualted and contains enough similarity or transferrability to be able to make accurate generalized predictions about the behavior of the simulated system. How and why that works with modern math methods is still somewhat mysterious. The most useful thing written about that so far is https://en.wikipedia.org/wiki/The_Unreasonable_Effectiveness...
Abstraction. You don't need to know what's going on at the quantum level because higher-level simulations can capture all the properties you care about.
This is a fundamentally flawed way of questionning, "knowing all the laws of physics" is probably impossible to achieve, but we still produce accurate predictions for a lot of phenomena. You can see that we do in fact "replicate" (predict would be a more proper term) things despite not knowing all the laws of physics: weather, movement of the stars, cooking time for a browned piece of bread...
I was replying to someone asking if it was "exact"… and no it isn't… You listing a bunch of simulations, that are notoriously not 100% correct proves my point, thanks.
The same way we can simulate the movements of the planets - it will never be exact, but the better we understand it, the more precisely we can simulate it
clearly depends on needed resolution and our ability to recreate the material structure. you don't have to understand much about wood to build a chair.
