Interesting thought - I think that there are challenges with the model, for example, RNA-only virii also have selection and bacteria do exhibit some methodologies like quorum sensing. But one could argue that eukaryotes do seem to have something special, and the idea of consciousness as a distribution from the cells is interesting and if true, would prevent mechanical AI from being conscious(or intelligent in the sense of seeking survival, being adaptive, etc.)
I'm not a huge fan of the idea of "simulating biological cells" since that is a lot of computation for the simulation, and I'm pretty sure that trying to do the universe simulator means that you run out of atoms before you fully simulate the universe.
I didn't either, until I saw the diagram in the "Agency" post. How can a material thing - including a "blob of protoplasm" (i.e., a human baby), be motivated? It has to be something unlearned (present *before* birth), at the cellular level (individual cells compete for everything), and unique to eukaryotic cells (bacteria don't have agency - they are machines in every way we think of them).
The only thing that meets all three criteria is DNA.
Even Alan Turing said something similar in his early work - I'll look it up, but I was unaware of it when I first "proposed the theory".
I'm open to alternatives, but I cannot think of any.
I like your idea but wouldn't something as simple as a reward function be a form of motivation? I mean, self-play can optimize for winning gamea like chess and it is "motivated" to make the best move, no?
Yes, but chess is a highly constrained problem space. What about open-ended goals like “survival”. A billion years of training data, hard coded into a protein, seems like a more reasonable solution.
They are, and they do! They just like to live in “mainframes” (animal and plant bodies) where they can hang out in Mission Control to pull the levers.
Interesting thought - I think that there are challenges with the model, for example, RNA-only virii also have selection and bacteria do exhibit some methodologies like quorum sensing. But one could argue that eukaryotes do seem to have something special, and the idea of consciousness as a distribution from the cells is interesting and if true, would prevent mechanical AI from being conscious(or intelligent in the sense of seeking survival, being adaptive, etc.)
I'm not a huge fan of the idea of "simulating biological cells" since that is a lot of computation for the simulation, and I'm pretty sure that trying to do the universe simulator means that you run out of atoms before you fully simulate the universe.
Why does agency require DNA? I don't see the reason.
I didn't either, until I saw the diagram in the "Agency" post. How can a material thing - including a "blob of protoplasm" (i.e., a human baby), be motivated? It has to be something unlearned (present *before* birth), at the cellular level (individual cells compete for everything), and unique to eukaryotic cells (bacteria don't have agency - they are machines in every way we think of them).
The only thing that meets all three criteria is DNA.
Even Alan Turing said something similar in his early work - I'll look it up, but I was unaware of it when I first "proposed the theory".
I'm open to alternatives, but I cannot think of any.
What's your take, Shon?
I like your idea but wouldn't something as simple as a reward function be a form of motivation? I mean, self-play can optimize for winning gamea like chess and it is "motivated" to make the best move, no?
Yes, but chess is a highly constrained problem space. What about open-ended goals like “survival”. A billion years of training data, hard coded into a protein, seems like a more reasonable solution.
But if true, wouldn't DNA molecules by themselves "be motivated" to replicate?