Re: Super intelligences

Robin Hanson (
Thu, 06 Aug 1998 14:21:29 -0700

Hal writes:
>> It seems to me that there is at least one striking difference worth
>> naming, between systems which have been designed and understood from the
>> ground up, and evolved systems which we have tweaked and adapted to our
>> purposes. Cars vs. Cows.
>By "evolved systems" do you mean strictly biological systems which have
>evolved over billions of years, or do you mean more general systems
>which use evolution as a design methodology? It may turn out that many
>artificial systems will be designed at least in part using evolutionary
>methods like genetic algorithms.

I mean both, though the effect is stronger for the former. Artificially evolved systems are similarly hard to disentangle, but evolution will likely be used to design small grain systems which are then combined into larger systems via explicit understood design.

>On the other hand, during the era in which we are biological entities
>there will be great value in understanding the tangled processes which
>produce life (and health), as well as finding ways to marry technology to
>these naturally evolved systems.

Great value, yes, but I think the problem is so hard that this won't translate into enough progress to change the overall characterization of such systems as mostly not understood and hard to flexibly make use of parts. By the time we can switch to fully designed systems, I think the temptation to do so will still be very strong.

Robin Hanson RWJF Health Policy Scholar, Sch. of Public Health 510-643-1884 140 Warren Hall, UC Berkeley, CA 94720-7360 FAX: 510-643-2627