From: Eliezer Yudkowsky (sentience@pobox.com)
Date: Sun Oct 24 2004 - 11:17:04 MDT
Eliezer Yudkowsky wrote:
> Ben Goertzel wrote:
>
>>> What exactly would you say is the difference between "philosophical"
>>> understanding and "mathematical and rigorous and definite"
>>> understanding?
>>
>> Eliezer, here is the difference.
>>
>> When you have a mathematically rigorous argument about some topic,
>> then the following holds: When you explain your assumptions and
>> arguments to ANY properly educated, non-insane listener, they then
>> will accept that your conclusions follow from your assumptions, with
>> close to 100% confidence.
>>
>> Philosophical arguments, on the other hand, have enough ambiguities
>> and holes in them that even when you explain your assumptions and
>> arguments to a properly educated, non-insane listener, they may not
>> fully agree that your conclusions follow from your assumptions.
>
> Would you characterize the definition you've just given as
> "philosophical" or "rigorous"? And do you think you could give a
> rigorous definition, if required?
Actually, let me amend that question. According to your given definition,
we can never make a "rigorous" statement about any real-world issue. The
empirical probability of a physicist's prediction is not certainty - it
can't be, not in an uncertain universe. Problem of induction and so on.
But physics is more than philosophy, and I would call it rigorous. And I
think we can both agree that for AI, what matters is not that the
conclusions follow with certainty from some set of assumptions, but that
the conclusions are actually correct. If someone says, "All Friendly AIs
are benevolent, all AIs based on reinforcement learning are Friendly,
therefore all AIs based on reinforcement learning are benevolent", the
conclusion follows from the premises, sort of, but the theory is neither
rigorous nor useful. So I think that the definition you offered may not
exactly match the distinction you had in mind, when you said that I was
doing philosophy.
So, again, do you think you could give a *rigorous* definition of the
distinction you had in mind when you worried that I was too "philosophical"?
-- Eliezer S. Yudkowsky http://intelligence.org/ Research Fellow, Singularity Institute for Artificial Intelligence
This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:49 MDT