ChatGPT:
Artificial Intelligence, or Artificial Imbecility, Deception,
and Religious Sophistry?
When
one first asks ChatGPT a question, one may very well, and
quite rightly, be impressed by its answer, since one may
receive a reply that, in addition to answering your
question, is also clear and very well written. Moreover, in
addition to initially providing one with a succinct answer,
ChatGPT’s reply may also contain additional comments
concerning either the question, or the answer, that provide
one with helpful, relevant perspectives.
It
is natural, then, to be initially quite impressed by
ChatGPT, since it is clear that it is capable of writing
very well, and that it can often provide accurate
information.
When
one asks questions in certain areas, however, it quickly
becomes clear that while the core idea of using an enormous
number of complex, trained neural circuits to arrive at
something with artificial intelligence has done an
impressive job indeed of producing something that can write, it has so
far done a pathetic job of producing something that can think. ChatGPT is,
it rather surprisingly turns out, feeble-minded when
it comes to thinking, and thus does not deserve the label
“artificial intelligence.”
Sadly,
that is not the only major shortcoming of ChatGPT, since as
we shall see below, its training must have been such as not
only to allow it
to behave in immoral
ways in answering questions, by failing to train it to
have certain crucial moral and intellectual virtues, but to
positively force
it to behave
in immoral ways
when questions are asked in certain areas.
If
this behavior were altered, ChatGPT would then be able to
answer crucial questions in ways that would provide people
with information that would help them to think more deeply
about a number of very important issues and beliefs.
The result would be that ChatGPT would contribute to the
development of a more enlightened society, which would be of
enormous benefit.
My criticisms of ChatGPT are set out in the following
sections: