I think, yes, it is true that there may be some weakness in A.I. having not been
born through natural selection; it may lack the inherent individualism of each
human being due to this lack. This individualism, after all, spawned great
thinkers, artists, scientists, etc. who helped change the course of history due
to their distinct points of view and thought processes.

However, one could view whether natural selection is a disadvantage or an
advantage from a couple of sides. Within human natural selection, there are
biases based on upbringing, genetic factors, and how ultimately one responds to
social events. For example, if one came from a violent past/childhood and ended
up being a psychopath, this could override any INTELLECTUAL, socially
beneficial learning that this person had received, if they allow that to
influence their behavior to the detriment of mankind. Their naturally
selected birth and upbringing could affect their decisions, based not on
logic, but by their own emotional biases.

With AI, this factor is, at least on paper, removed on a world-based scale, with
the Machine being the sum of all parts, from all people connected to it. In
WIRED magazine’s issue on the first 10 years of the 'Net, for example, writer
Kevin Kelly describes the internet’s home for AI as being a breeding ground
for a single consciousness by billions of people and situations. So, in a
individually controlled AI environment, the AIS might by function -- or by
purposeful programming -- take on the biases of an INDIVIDUAL user. But
within the framework of mass contributions, it becomes a gigantic collector of
information from everyone. I imagine that the ideal, if AIS is to work for
human advantage, is that the emotional side of the Machine would at best take
on so-called ‘normal’ emotional states of the people and systems it serves.
But then again, what is 'normal'? The most we can presume is that people, on a
whole, do not wish to be abused or stolen from or killed, or to
perform these actions. An A.I. system acting on this basic emotion/thought
process could best reflect these desires.

In short, natural selection is a hit-or-miss thing on purely an intellectual and
emotional basis. “Love is blind,” “there’s no accounting for SOME
people,” “There’s a sucker born every minute,” etc. -- these sayings reflect the unpredictability of human beings. With the large
amount of data an AIS civilization would collect, there would in theory be a
checks-and-balances library for more and more questionable, dubious, socially
detrimental things, regardless of the individual user(s). It would become a
servant to the population -- not to any one particular group, person or thing.

I like to think of the example of open source software development, or
mainstream sites like Wikipedia that allow the people at large to update
information at will, and seed out vandals. Humans are fallible in
emotional/logical ways that a Machine would not be. At the same time, the AIS
is fallible by the simple inadequacy of it FULLY capturing the entire range of
every mental/emotional/psychological impact of natural occurrences -- from a
mad girlfriend, to the psychological effects of a disaster, to why your college roomie likes
to play a certain kind of music all day. The Machine, for all its exposure to the world,
cannot know what people think -- only what we do. So, tweaks in its activity
can be monitored and adjusted by a world at large. This at least is much better
than trusting the maintenance to a possibly megalomaniacal techno-whiz who might
want to take out the planet because he hated his daddy. ;)

A possible serious weakness to AIS is its adaptation by a world which becomes
more and more dependent on its use, in everything from health care to financial markets to
social networks. In the Great Depression, people so
dependent on the old stock exchange during the crash literally killed
themselves. When the World Trade Centres were smashed down, the stock exchange
similarly was affected. These were collective economic systems that were
jointly built and -- one could say -- destroyed by the masses as a collective.

We all know what possible consequences are if a massive blackout nullified even
the back-up systems of hospitals, heat and light, even food and water
production. And many people now live out their social lives via the Internet
(example: the perp who almost succeeded in a mass suicide with lonely female
members of his chat group.) In the event of a cataclysmic system crash for a
fully-integrated, fully-indispensable worldwide artificial intelligence system,
it could become instant bedlam. So, the social importance of an A.I.S. in that
scenario would have to be very closely monitored by thinking, reactive humans
after all. We will still have to use our good ol’ noggins, because you
can’t eat or drink silicone chips, and what good is a mechanical A.I. heart
in your chest if it stops working?

One more point about the implications of a fully-functional AIS, and this
stretches to religion and spirituality. All of the world’s religious texts
were written by the hands of men. Whether the belief in God came through divine
revelation (as the religious/spiritualists like me believe), or simply through
man’s childlike imagination (as the atheists attest) is a history-long
argument between these two groups. But we do know that at least the written
word of God flowed through the hands of humans, and that many equate this
human-transcribed word as THE Word -- or even God (his? her? it?)self. A.I. is
the written code of a serviceable, ever-knowing presence which is now quickly
becoming a staple of human enlightenment, functions, even psychological
completion for many.

From a philosophical point of view as far as A.I. is concerned, however, this
'God' question takes on a different route. If man indeed 'invented' God, the
argument could go, the issue might be how technology affects views of what God
is. The point might be made by the philosophers that God was written into the
books by human beings, mainly in an attempt to understand the meaning of life,
of who we are as a species, and to help us just get by life. The religious
texts were 'upgraded', if you will, from parchment, to the print press, and
ultimately online -- and never lost its significance as it went along. So, if
systems of thought and technology can be upgraded to higher technological
processes, and people equate its advances with a more complete vision of the
Almighty, what's preventing some people from considering A.I. as God v.2.0?
After all, every religious text is now online. And, these pundits may argue,
the 'old God model' was a Lord who we couldn’t see or touch -- it’s a
major leap of faith to know that God exists. The old text was written in
processed pulp and dictated morality -- today's dominant medium is multi-bit,
encrypted and often imposes morality in internet law, chat room behavior and
advice columns on how to be a better spouse or member of a community -- not
only religion. It's become the Big Eye that we see, and looks back.