Re: What best evidence for fast AI?

From: Stathis Papaioannou (
Date: Mon Nov 12 2007 - 16:40:50 MST

On 13/11/2007, Peter C. McCluskey <> wrote:
> (Eliezer S. Yudkowsky) writes:
> >I would begin by asking if there was ever, in the whole history of
> >technology, a single case where someone *first* duplicated a desirable
> >effect by emulating biology at a lower level of organization, without
> >understanding the principles of that effect's production from that low
> Here are a few innovations that happened in a way that resembles uploading
> more than it resembles designing AI based on the principles of intelligence:
> - producing carrots and strawberries that are an order of magnitude
> larger than their wild ancestors. This process appears to have used methods
> of evolution before evolution was understood. Alternatives such as writing
> new genomes from scratch or telling a molecular assembler where to put the
> atoms still look hard.
> - aspirin was created by tweaking an herbal medicine before we began to
> understand how pain surpression works.
> - inoculation as a means of reducing smallpox deaths appears to have
> started about two millenia before there was a clear understanding of
> how it works.

I don't think those are examples of what Eliezer was asking for, but
rather the opposite. The equivalent of mind uploading without knowing
how intelligence works would be, with the first example, producing
larger vegetables by directly altering their DNA without understanding
how genetics works. In general, if you understand the low level
structure of a machine well enough to reproduce the machine, you will
probably also understand how that machine works. Imagine scientists
from the 19th century reverse engineering a modern computer but not
understanding how computers work.

Stathis Papaioannou

This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:01:00 MDT