5 ms·
I think it's a bit of hubris to demand we explain how LLMs can be as intelligent as they are when we barely understand how the ball of meat inside our skulls ca
by mabbo 3y ago
I think it's a bit of hubris to demand we explain how LLMs can be as intelligent as they are when we barely understand how the ball of meat inside our skulls can be either.
We know that neural networks can simulate any function, given enough parameters. Maybe we've simply found the number of parameters needed to simulate the function of "human level intelligence".
That should humble us, to know that there is some number which we are no more complex than.
- gfody 3y ago> there is some number which we are no more complex than I bet that number is 808017424794512875886459904961710757005754368000000000
- hammyhavoc 3y ago> Maybe we've simply found the number of parameters needed to simulate the function of "human level intelligence". Have we? That seems like a huge reach.
- the_optimist 3y agoWhat is the meaning of your philosophical approach? As in, what is the proposed consequence of such hubris or humility?
- peter303 3y agoThe physical human brain has 86 billion neurons with up to 10,000 connections each. However they are somewhat structured, e.g. the seven layers of the cerebral cortex, reducing the arbitrariness of connections. So say the connectome has 100 trillion parameters. The bain also has a bunch of other cells that dont appear to encode information. They may supply structural scaffolding and chemicals to mediate neuron function.
- peter303 3y agoThe physical human brain has 86 billion neurons with up to 10,000 connections each. However they are somewhat structured (e.g. the seven layers of the cerebral cortex segregated into vertical silos a few hundred cells wide) reducing the arbitrariness of connections. So say the connectome has 100 trillion parameters. The brain also has a bunch of other cells that dont appear to encode information. They may supply structural scaffolding and chemicals to mediate neuron function.