Re: [sl4] simplistic models of capability growth

From: Petter Wingren-Rasmussen (petterwr@gmail.com)
Date: Mon Feb 09 2009 - 11:26:19 MST


Defining things is a very good start :)

On Thu, Feb 5, 2009 at 2:27 AM, Johnicholas Hines
<johnicholas.hines@gmail.com> wrote:

> 6. This last one takes a bit of motivation. Suppose the system in
> question has two parts, one which is modifiable by the system and one
> which is not. The total capability of the system is a combination of
> the two parts. But the combination is not a simple addition or
> multiplication of figures of merit. Rather, the total capability of
> the system is measured in "speed", and each "elementary cycle" needs
> to pass through both the modifiable part and the nonmodifiable part.
> There is one endogenous variable "modifiable part's speed" (m).
> There are two parameters, "modifiable part's improvement per
> capability" (mpc), and "nonmodifiable part's speed" (n).
> There are two equations:
> The total capability of the system, the speed, is the inverse of the
> sum of the inverses of the modifiable and nonmodifiable parts' speeds
> (c = 1/((1/n)+(1/m))).
> The time derivative of the modifiable part's speed is proportional to
> the total capability (D[m]=mpc * c).

If I understand this correctly it means that any non-modifiable part
of an AI will sooner or later be the limiting factor for its growth.
This leads to a very large risk in the long run of a competing system
without a non-modifiable part to become more efficient/powerful than
the one with non-modifiable parts.
For me the conclusion is clear:
Any AI with a hard-coded law for "being nice","prevent murders" or any
other non-modifiable procedure will with great probability be overrun
by a system without those hardwritten rules.



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