Page 66 - Artificial Intelligence in the Age of Neural Networks and Brain Computing

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

Homo sapiens have two forms of intelligence associated with each side of the
brain: Emotional Intelligence Quotient (EIQ) (intuition, empathy, love, artistic sen-
sibility) [1] and Rational/Logical-IQ (LIQ) (language, symbolic manipulation, fore-
casting/planning, learning). Deep learning algorithms have shown it may be possible
to endow these characteristics to machines to produce some type of what humans
might term “smarts.” However, without the wisdom of EIQ, Automation technology
(AT) cannot be endowed with LIQ, let alone be sensibly intelligent. The earliest
form of the beginnings of mechanizing of forecasting/learning is a control mecha-
nism known as feedback. Historically speaking, feedback control was instituted
millennia ago by the Greek engineer Ctesibius (285 BCe222 BC) in a device known
as a water clock that regulated water levels to indicate the passage of time. The heat
regulation mechanism for a chicken coop, the thermostat, was invented by Cornelius
Drebbel in 1620. An automatic flour mill was developed by Oliver Evans in 1785 by
adding a fantail to keep the face of the windmill pointing into the wind. James Watt
in 1788 invented the centrifugal governor to control the steam engine; it was latter
mathematically analyzed by James Maxwell forming the basis for industrial control
theory which lead to first Gen completely automated industrial process. Further me-
chanical developments such as servos, open loop control, autopilots, and analog me-
chanical machines (computers) followed. All of these concepts plus on-loop, close
loop, discrete, proportionaleintegralederivative controller (PID), sequential, logic
sequence, could be implemented in electronic devices such as electronic circuits
and electronic computational devices at various scales and degrees of complexity.
Feedback as a control mechanism has entered other fields with its own unique
ontology for distinct fields. For example, in modern governmental economics, ac-
cording to former Federal Reserve Chairman Alan Greenspan in his retirement
speech, the US inflation has been controlled mainly due the feedback from Chinese
providing high-quality goods at low cost that keeps US dollar value stable despite
large trade deficits. Analogously, a disruptive technology in the next 10 years would
be thousands of DAV trucks carrying consumer cargo-box goods to their chain stores
by such companies as Wal-Mart, CVS, Safeway changing the traditional drivers in
the loop delivery mechanism with an automation intelligence, for example.
Likewise, Artificial Rule-Based IntelligencedAI for shortdbegun by Marvin
Minsky (MIT) c.1970, was the first introduced rule-based system which we refer
to as first Gen AI. AI has been largely viewed as a failure in the late 1980s:
“They promised us a child prodigy; they delivered an idiot savant” (because a rule
is made to be broken in truly intelligent behavior). Frank Rosenblatt (Cornell) intro-
duced the analog concept of neural networks of human visual system. Because of the
limitation of size-weight-speed-power (SWSP), the early analog device was made of
20 20 units of photodetectors as the input layers, which were interconnected by
512 stepping motors threshold unit’s and an output feature layer. Such a primitive
computational artificial neural network (ANN) was returned from Office of Naval
Research (ONR), after Minsky’s criticism that it cannot even compute the EX-OR
truth table; but it was kept as S&T landmark at Washington, DC, Smithsonian
Museum. In early 1990s efforts to implement digital ANN using traditional digital