SV: Capacity of the brain

Will Dwinnell predictor at compuserve.com
Thu Sep 2 09:50:55 EST 1999


Russell Wallace wrote:
"In round numbers (assuming our current understanding of how neurons
work is more or less correct), here are the raw figures for the human
brain:

Storage capacity: 10^15 bytes
Processing speed: 10^18 calculations per second

A couple of other data points: If Moore's Law continues, supercomputers
will duplicate this processing speed around 2030, with desktop machines
achieving it by around 2045.  (The storage capacity will be duplicated
and surpassed well before then.)  It obviously needs to be borne in mind
that reaching this level of hardware performance is a necessary but not
sufficient condition for creating a human-level AI."

This analysis strikes me as vastly simplistic.  I suppose that at some
very low, information-theoretic level, the information processing
provided by both of these systems is the same, they differ markedly in
higher-level processing.  Since computers and biological nervous systems
perform many different types of tasks, it would seem self-defeating to
me to reduce analysis of their performance to a simple scale. 
Computers, for instance, have already far and away surpassed the
abilities of people for many tasks.  I would think that one would be
very hard-pressed to defend such specific metrics as those given above
in all but the most broad interpretation.  Human brains do not act as
random access memories.

Will Dwinnell
predictor at compuserve.com



More information about the Neur-sci mailing list