From: William Pearson (wil.pearson@gmail.com)
Date: Thu Oct 25 2007 - 09:45:09 MDT
On 25/10/2007, Rolf Nelson <rolf.h.d.nelson@gmail.com> wrote:
> Bayesian reasoning involves determining a prior probability "before you've
> considered the data", and then getting a final posterior probability by
> shifting based on the data you've collected. It can be applied if you have
> 2000 data points, or if you have 2 data points, or even if you have 0 data
> points (in which case the posterior probability remains trivially equal to
> the prior probability.)
Thinking that it is probable that humans will create AI, is not part
of humanities prior, we have all acquired that belief in one way or
another. My question was partially a test of the bayesian view of
intelligent agents, as well as an honest inquiry because I do like
statistics and backing up my beliefs.
If there was a belief that had been acquired (so not in their prior)
by a group of people, that I could not find a way to collect evidence
that should have adjusted their prior, I would have had to have
concluded one of three things.
1) It is acceptable for intelligences to acquire beliefs in the high
probability of events that there has been no statistical evidence for.
1.1) Which would mean that the view that, bayesian reasoning is all we
need to explain what intelligences should do, is lacking.
2) They were thinking incorrectly and shouldn't have held that belief.
3) There is other evidence I am missing.
It turned out to be number 3, in this case. However, in order to be
able to say there is evidence for AI to be probable, I would need to
do some actual stats on how well the success rate of inventors is
correlated and whether you can justify predictions about future
inventions of certain types (bio-mimetic ones for example) from past
successes. But at least you can hypothesise that society as a whole
has done something like that in an unconscious fashion. Our beliefs
about AI being probable in this case being received wisdom, of the
good sort.
Will Pearson
This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:58 MDT