151–155 of 155 posts
Earlier quoted context omitted.
The octopus is a counterpoint to this, being very intelligent but absolutely solitary.
But it has excellent hands. Humans may be unique in having both complex social structure as well has hands.
"The evolution from squid to octopus is compatible with a suite of genes inserted by extraterrestrial viruses. An alternative extraterrestrial scenario discussed is that a population of cryopreserved octopus embryos soft-landed en mass from space 275 million years ago."
https://www.sciencedirect.com/science/article/pii/S007961071...
Earlier quoted context omitted.
Well, a couple of things start to make sense about the kludginess, if you start with the single-cellular origin, and then pile up layers of specialized tissue, with some cells permanently buried within the organism, never to be exposed to an external environment. The first part is that brains are just plain weird, and perhaps one of the weirdest aspected of any multicellular organism. More than half of all known life…
Crabs and lobsters don't use external digestion but eat their food not terribly different from what we do. So it looks like spiders evolved their system later.
In all invertebrates, "brains" (to the extent that they're present) are ring ganglia, with the gut going through the center. So arguably they all face that tension between brain development and the ability to eat. Vertebrates, conversely, have ~parallel guts and central nervous systems, so there's no conflict.
The thing with octopuses is that their ring ganglia are mode widely spaced. Having no exoskeleton helps with that, I guess.
Earlier quoted context omitted.
But it has excellent hands. Humans may be unique in having both complex social structure as well has hands.
Arguably elephants fall somewhat into this category...
I guess Niven and Pournelle had the same idea for Footfall.
Earlier quoted context omitted.
It's not luck, it's time. Evolution occurs so extraordinarily slowly, it's hard to comprehend. People are trying to create intelligence a million times faster than evolution did.
Well that's the usual view but it's based on random mutations so there must be some luck involved. A change of the make this muscle a bit bigger type may involve the equivalent of one bit flipping and so happen all the time but the mechanisms of intelligence may involve a much harder sequence to come up with. It's hard enough to program neural networks using code and tensor flow. Imagine having to do it by giving the…
Computers struggle to handle a single genome, let alone mix and match multiple. And even when they do, it's a single organism. In nature, there are thousand's of organisms mixing and matching, thriving and dying all the time, spread over hundreds of millions of years.
No computer's bandwidth could come close to that.
But I have a deeper discomfort with your question. It assumes that the neural network / tensor-flow model we use with computers approximates a human's. I'm not terribly convinced of that.
Sure, human neuron's have are individually approximated by neural network neurons. But concepts such as "instinct" are not captured well by that model.
It's weird that a newborn baby has seen far fewer colors, and patterns than a well trained neural network, yet seems to know far more. It signifies we're missing something pretty central to the model.