I wonder how accurate would simulations need to be and what aspects of a neuron are fundamental to intelligence and what aspects are just implementation details and hacks made by evolution to keep this squishy glucose powered machine as efficient as possible.
I mean, cells made of fat, water and proteins must come with limitations that are completely different from the limitations of cells made of silicon, fiberglass and metals.
So if we are only trying to replicate the fundamental properties of a neuron and not its inner workings, what do we scrap, what do we keep, and what can we change to make it even better for running on computers?