[ExI] Smallest human-equivalent device
Anders Sandberg
anders at aleph.se
Fri Oct 11 08:49:30 UTC 2013
> ### I am surprised that the estimate is so, well, bulky. 1 liter
> volume is very close to actual human brain volume and I simply don't
> believe our brain is anywhere near the limits of miniaturization...Rafal
The typical estimates work by assuming a scale, counting the number of
things on the scale, estimating the computational requirements per
thing, a conversion factor from computation to physical resources in an
artificial system, and then multiplying. There are plenty of
arbitrariness here, a bit like using the Drake equation. Especially the
scale choice can totally change things without looking like a massive
assumption.
Personally I suspect a straight mapping brain-to-solidstate will have to
be bulky, while a functional mapping can be very small.
On 2013-10-11 04:46, spike wrote:
> In all these kinds of calculations we are locked into a paradigm that
> was handed to us by evolution: brains exist as individual units. We
> need to think of ways in which silicon based intelligence can work
> together in teams far more effectively than we carbon units can ever do.
If you have program A and program B, with goal 1 and 2 respectively,
there can exist a program AB that achieves both goals in some optimal
tradeoff and saves resources by combining A and B (think shared
libraries and no risk for conflict). So under some conditions it is
rational to merge software into bigger systems. Humans typically balk at
this because their goals include existing as themselves, and these
lexical goals are incompatible with the merging. But other software will
disagree. And the uploads in my big data center will not care that they
are sharing underlying hardware and software as long as their data and
processes are kept in separate virtual machines.
Biobrains are messy: creating good interfaces to them is hard.
--
Dr Anders Sandberg
Future of Humanity Institute
Oxford Martin School
Oxford University
More information about the extropy-chat
mailing list