This is an interesting post and I agree with it. Also, modeling a calculator at the transistor level is a useful analogy.
As for the discussion on "286s or 1995 home computers" these computers seem capable of containing in memory both the DNA description of a simple Bacteria like Pelagibacter ubique (1,308,759 base pairs) as well as simple model of all the atoms in the largest protein that Pelagibacter ubique contains. So this level of computer might be sufficient to bootstrap to a more powerful computer by designing the computational and energy gathering technology and then building new components. Conversely, it would be very hard to prove that a 1995 home computer is incapable of becoming a seed AGI. I have expanded my thoughts on this in a draft paper at https://www.researchgate.net/publication/388398902_Memory_and_FLOPS_Hardware_Limits_to_Prevent_AGI
This is an interesting post and I agree with it. Also, modeling a calculator at the transistor level is a useful analogy.
As for the discussion on "286s or 1995 home computers" these computers seem capable of containing in memory both the DNA description of a simple Bacteria like Pelagibacter ubique (1,308,759 base pairs) as well as simple model of all the atoms in the largest protein that Pelagibacter ubique contains. So this level of computer might be sufficient to bootstrap to a more powerful computer by designing the computational and energy gathering technology and then building new components. Conversely, it would be very hard to prove that a 1995 home computer is incapable of becoming a seed AGI. I have expanded my thoughts on this in a draft paper at https://www.researchgate.net/publication/388398902_Memory_and_FLOPS_Hardware_Limits_to_Prevent_AGI