Re: Ethical experimentation on AIs

From: Lee Corbin (
Date: Sun Oct 21 2007 - 01:40:10 MDT

Stathis writes

>> Yes, there is *very* good reason to suppose that the death throes
>> of a fly exposed to insecticide are not associated with great pain.
>> Evolution does not find it expedient to waste resources for no
>> purpose; no complex behavior or module ever arose unless
>> it aided the survival of the creature (its genes)...
>> The amount of pain a creature feels is in direct proportion to
>> the degree of control it has over its environment, and the way that
>> such control has in past evolutionary history paid for itself.
> Your argument may be valid for trees, but I don't see why it should be
> valid for flies. There are some environmental factors which are good
> for flies which they actively seek and others which are bad for flies
> which they actively avoid.

Yes, you're quite right about that.

> We can't know for sure what flies feel but a pleasure/pain response
> would be an effective evolutionary strategy to bring about the right
> sort of behaviour.

But the pain that a fly experiences would only be that sufficient to "get
its attention", as it were: whatever minimal amount that would be
required for it to attempt to fly away. My ignorance of fly-behavior
prevents me from speculating on how much, exactly, you can change
a fly's behavior with training. But my point is that the more automatic
the behavior, the less self-awareness is necessary and the less good
that highly evolved pain circuits would do.

> If anything, you could make a case that while it might be appropriate
> for intelligent beings such as us to have a mechanism for putting up
> with pain when we have some ultimate goal in mind,

Exactly! That's why it's criminal for medical practitioners to withhold
pain medication, and even more criminal for philosophers to try to
rationalize severe pain in folks. (Of course you, far more than I, know
about whatever actual tradeoffs for patient well-being exist
when it comes to the application of pain medication.)

> It would pay for a fly to suffer terrible, unremitting pain until the
> noxious stimulus was removed.

Instead, I would say that *we* are capable of experiencing terrible
unremitting pain precisely because we have so many higher-level
means of overriding it. Often, as humans, we really do have to
experience a little or a lot of physical discomfort to achieve a more
subtle purpose. That's why the pain is terrible---it's evolution's way of
saying "Are you really sure about this?", if you were to try to override it,
or to attempt to deter you from making plans that would damage your

But a fly has no abilities to override its basic instincts. Hence, there
is no purpose to be served by the its experiencing severe pain, and
so nature does not provide it the capabililty.


This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:58 MDT