'What is your name?' 'Ken Clements.' 'Do you deny having written the
following?':
> This may be a good time for people on this list to think about the Good v.
> Bad memeplex. We need this to provide a foundation for the discussion of
> the development of future technology, and is directly applicable to the
> question of how AI devices making decisions will impact *our* future.
As far as I can tell, the lesson of this story appears to be skepticism
about our capacity to know the total consequences of a given action. Of
course, we take skepticism in moderation. Too much or too little is, dare
I say it, Bad.
Good is a useful word, because there are some things which we ought to
pursue; we call them goods. (Though, as in this story, some things which
we ought to pursue land in our laps.)
Here's my current favorite introduction to ethics for people who are
committed to rationality.
---Here's an interesting fact: You're stuck. You have to make choices, both about what you're going to believe, as well as about what you're going to do. Even if you choose to just sit there, that's still a choice. You can't escape this.
Uncoincidentally, we have a handy vocabulary for beliefs and actions in terms of their rationality. Believing strongly that A in the face of strong evidence that not-A is irrational belief. Walking across a busy freeway blindfolded (under normal circumstances) is an irrational action. I assert that we can figure out both beliefs and actions rationally. Call the rational set of actions "ethics."
You can commit yourself to the claim that we cannot ascertain what constitutes either a rational belief, but that certainly appears to be against rationality (since, on this view, no beliefs or actions are rational). Similarly, you could claim that we can figure out rational beliefs but not rational actions. However, the overlap there is quite strong; if you grant me the one, it's hard to see how you could fail to grant me the other. Believing, after all, is an action, so if we don't have rational actions, then no subset of actions are rational, including beliefs. So if there are no rational actions, there are no rational beliefs. So if there are rational beliefs, there are rational actions.
So how are we to go about this project? The same way we go about figuring out which axioms are correct. This process is not deductive, though it can be inferential. It is a process of maximizing the coherence of our axioms with one another, which includes our beliefs about what the world is like and which actions are paradigmatically rational and which aren't. Or, more generally, you think about it a lot, talk about it, consider the possibilities, and see what you can come up with.
---
-Dan
-unless you love someone- -nothing else makes any sense- e.e. cummings
This archive was generated by hypermail 2b29 : Thu Jul 27 2000 - 14:06:21 MDT