Re: When does it pay to play (lottery)?

From: Harvey Newstrom (mail@HarveyNewstrom.com)
Date: Sun Jan 23 2005 - 13:14:59 MST


On Jan 23, 2005, at 12:23 PM, David Clark wrote:
> I thought the points made by Harvey Newstrom about taking an
> engineering
> approach was very good, however, engineering works best when you know
> your
> tools and your requirements. How do you use an engineering approach
> when
> you don't know what tools to use, what exactly the problem really is,
> or
> even what approach to take to solve the problem? With an AGI,
> *everything*
> is an unknown. Engineering works with things that are known.

Security engineering does not work with things that are known. Hackers
try creative things. New exploits are discovered all the time.
Unexpected interactions occur between systems and over the network.
Unexpected bugs exist all over. Technology can fail in
catastrophically unexpected ways.

Come to think of it, there are a lot of fields that work with the
unknown. Safety doesn't work with the known. Policing or war actions
don't work with the known. Exploring space doesn't work with the
known. Medicine and healthcare are often unpredictable. Most of
science is exploring the unknown. This is not a unique problem to AI.

What you describe as engineering the known only applies to
mass-produced widgets on an assembly line, or standard architecture
techniques using known technology. All scientific developments and
research face the same difficulties with trying to approach the vast
unknown and develop the state of the art until it becomes known. This
is standard science. It can be done safely.

Witness the Manhattan Project, the Apollo Missions, and AIDS
treatments. I would argue that standard security, safety and
precautionary measures already exist in most technology fields.
Combining those safety protocols from environmental hazards, medical
testing, psychology experiments, computer security, and policing enemy
combatants should cover the many of the same unknowns that an AI might
present to us.

I'm not saying there aren't unique unknowns, but the challenges
presented by AI are not entirely outside the experience of what
scientists have faced before.

--
Harvey Newstrom <HarveyNewstrom.com>
CISSP, ISSAP, ISSMP, CISA, CISM, IAM, IBMCP, GSEC


This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:50 MDT