Page 68 - Artificial Intelligence in the Age of Neural Networks and Brain Computing

56 CHAPTER 3 Third Gen AI as Human Experience Based Expert Systems

for example, “Young” and “Beauty” which are ill-defined; they are all open sets of
possibilities. They cannot be normalized as the unit probability, but the open sets of
possibilities are powerful human thinking. It may be referred to as FMF, as intro-
duced by Lotfi Zadeh and Walter Freeman (Berkeley). However, when the Boolean
logic is based on set theoretic concepts of “union and intersection,” the Young with
the Beautiful results in a sharp increase in the meaning that all humans would under-
stand. A truism “A rule is made to be broken” indicates truly intelligent human
behavior if the result is deemed to be sensible. All these are not brand new, but
the community of automation computer scientists (ACS) is not accustomed to
modeling the range of reasonable human behaviors. Any automation environment
that includes humans must consider “what if the human does?” which the analog
possibility thinking is. Otherwise, the manemachine interface will have the mis-
matching challenge always occurring. A man might accidently cross the pedestrian
zebra line despite the red light.
Many scientists believe that one of the most important problems facing
21st-century science is to understand the human brain: how the brain’s functionality
gives rise to human intelligence. This is what we refer to throughout this document
as natural intelligence (NI), because of Darwinian mechanism of survival of the
fittest what might be termed “survivor natural wisdom.” Within this broad subject
area of the brain lies the problem of machine intelligence. Alan Turing raised the
problem of distinguishing between natural intelligence and machine intelligence.
This problem is now referred to as the imitation game. If one were interacting
with an intelligence that could be human or machine, how does one distinguish
between a natural intelligence from machine intelligence? By positing an inquisitor
who was allowed to ask as many questions as one found necessary to test the intel-
ligence of the subject, and if one could not determine whether the intelligence was
human or machine, then one could not distinguish an imitation from the real thing,
so that there is no difference between the two. One could substitute creativity for
intelligence and draw the same conclusion. This game is termed the Turing Test
and the question it raises is at the heart of deep learning revolution. It is one of
our central theses that casting interactions in game theoretic terms helps bring in
the concept of learning with teacher and without a teacher into the forefront of
understanding “intelligence” in natural learning, as well as artificial learning. Just
as learning is a human attribute that we wish machines to exhibit, we also wish
them to exhibit creativity as well. If intelligence is viewed as something that can
be imitated by a computer, then surely creativity should surely be considered as a
human attribute that can be imitated, as well as the Cardinal Rule: “Thou shall do
no harm to mankind” with all possible different initial boundary conditions.

2. THIRD GEN AI

This third Gen AI or second Gen AI may be called general artificial intelligence
(GAI) sequence slightly modified from the original by Lex Fridman of MIT, lectured