March/April 2019 I 39
Therefore, forward-thinking
ICT professionals no longer view
AI as merely algorithms and
software in support of smart
products and services but rather
as a vital functional component
of today’s and tomorrow’s ICT
networks and applications.
CLEARING THE CONFUSION
BETWEEN AI, COGNITIVE
COMPUTING, AND
AUGMENTED INTELLIGENCE
Amid the commercial media hype
for AI, true AI is suffering from a
case of mistaken identity. In the
article, Artificial Intelligence Has
Become Meaningless, it is argued that
AI “has been hijacked by companies
wanting to make their software
algorithms sound smarter than they
really are. Chat bots are often
classed as AIs, for example, when
they are mostly glorified phone
trees, or else clever, automated
Mad Lib-type programming.”3 For
example, upon close inspection,
Facebook’s so-called AI that detects
suicidal thoughts posted to its
platform is little more than a
pattern-matching filter that flags
posts for human community
managers, while Google’s so-called
Perspective machine learning
algorithm to identify toxic online
comments can be easily fooled
by simple typos.4
Further blurring the definition
of AI is the confusion between
identifying a cognitive computing
system and an AI system. Both are
needed in supercomputing and big
data, but each has a unique purpose,
task, and goal. Essentially, cognitive
computing is designed to “assist”
human decision-making, while
AI is designed to replace it. Cognitive
computing solves problems the way
people do by analyzing, reasoning,
and remembering. Conversely, AI,
on its own, can autonomously make
decisions in isolation without
human intervention using
algorithms and processes that are
not necessarily human-like in
cognition, and it does not mimic
human-like thought in decision
making as does cognitive
computing. The ability for AI
to make decisions without human
intervention, for example, is
paramount for the many unmanned
micro data centers that will be
installed as edge computing evolves,
as well as IIoT, smart cities and 5G.
The idea that AI is designed to
replace human thought—coupled
with the now infamous tweet
of controversial tech leader Elon
Musk who proclaimed that AI
Artificial Intelligence (AI), a very
old technology dating prior to the
1920s and with 40 year old roots in
ICT, is central to making Klaus’
prediction a reality. So much so,
Intel has invested well over $1
billion in businesses that are
advancing the field of AI, updated
its total available market estimate
for all AI silicon (i.e., servers,
accelerators, memory, networking,
storage) to $200 billion by 2022,
and launched an aggressive strategy
to compete with NVIDIA for AI,
AMD for data center graphics
processing units, and the many
other AI players vying for market
dominance.2 Moreover, recent
research forecasts predominantly
agree that global artificial
intelligence uses will grow
at a compound annual growth rate
(CAGR) between 50 to 63 percent
over the next three years.
Though AI has subtly impacted
ICT throughout the years, it has
been a subject far removed from
commonplace ICT discussions;
it was a technology out of the realm
of Layer 1 connectivity and was
best left to IT, data processors,
coders and mathematical wizards.
However, as the ICT industry
is transitioning into the IoT world
of network convergence and IP
device-driven global connectivity,
ICT designers and installers should
become familiar with AI’s general
capabilities, misconceptions,
trends, and its most lucrative uses
in order to recommend, design
and build the most innovative
and efficient networks for their
enterprise customers.
Each and every decade that followed
the 1920s to the present day promised
that an autonomous vehicle would soon
be ready for commercialization.