IEEE Robotics & Automation Magazine - September 2019 - 96

individual rationally deals with envi-
ronmental changes throughout her/
his lifespan. It is a definition that coin-
cides with Hawking's, as both consider
adaptation to be the main attribute of
intelligence. Adaptation implies the
ability to deal with perplexing situa-
tions and respond quickly and suc-
cessfully to new ones. Other signs of
intelligence include
● making sense out of ambiguous or
contradictory messages
● using reason to solve problems
● understanding and inferring in ordi-
nary, rational ways
● applying knowledge to manipulate
the environment
● recognizing the relative importance
of different elements in a situation
● learning and understanding from
experience [2].
The key attributes of AI may be summa-
rized as the ability to learn, reason, and
self-correct [3].
AI aims to mimic or reverse-engi-
neer biological intelligence to build
machines that are able to function and
interact autonomously within unstruc-
tured, dynamic, and sometimes par-
tially observable environments. This
usually involves borrowing characteris-
tics from human intelligence, such as
situational awareness, decision making,
problem solving, and learning from the
environment and adapting to changes
there. AI encompasses many fields,
including robotics, image recognition,
speech recognition, natural language
understanding, and machine learning
(ML). According to most experts, AI is
considered to consist of algorithms (the
software only); once its capabilities are
embodied, we may call it robotics or
something else. Given AI's popularity,
its name can be abused. In 2017, think-
ers from the IEEE Standards Associa-
tion (IEEE-SA) and the Massachusetts
Institute of Technology (MIT) Media
Lab founded the Global Council on
Extended Intelligence. That body
released a report, "The Case for Extend-
ed Intelligence" [4], that discussed
machine intelligence in terms of
capability in an effort to avoid the con-
fusion surrounding the anthropomor-
phizing of technology, in general. It also
96

*

IEEE ROBOTICS & AUTOMATION MAGAZINE

*

discussed the limits of AI-based tech-
nology for practical purposes in com-
putational systems.
AI and ML form the foundation of
many of the successful disruptive
industries we are witnessing, including
autonomous driving, cognitive health
care, assistive technologies, precision
agriculture, home intelligence, and
Industry 4.0.
It is important to stress that AI takes
many forms and may support research,
development, and innovation projects
in various ways, sometimes completely
hidden, as an underlying core technolo-
gy. It is believed that its uses and bene-
fits will be tremendously greater than
what is outlined in this article. A recent
report from McKinsey collected more
than 160 use cases in which AI directly
served society, a result that was well in
line with the SDGs [5].
AI Evolution
AI is an evolving technology that has
been present for many decades and was
born of a human longing to support
societal needs through intelligent
resources. AI development can be traced
to the classical philosophers and their
efforts to model human thinking as a
system of symbols, which led to "con-
nectionism" as a process of thinking.
One should differentiate between two
AI approaches: top-down (white box or
generative) and bottom-up (black box,
discriminative, or data driven). Tradi-
tional AI techniques follow top-down,
deductive reasoning and begin by
formulating abstract, wide-ranging
hypotheses about the world from preex-
isting knowledge and very few exam-
ples. They make predictions about
what the data should look like if those
hypotheses are correct and revise their
conjectures depending on the outcome
of those forecasts [6] and upon receiving
more observations.
Modern AI, or computational intel-
ligence, techniques are characterized as
bottom-up, inductive reasoning (work-
ing on numeric data to infer symbols),
and they try to extract patterns from
raw data (thousands or millions
of ex amples). They have powered
different products and services by

SEPTEMBER 2019

endowing them with the ability to dis-
cover and recognize complex pat-
terns and trends (classes, clusters,
and anomalies) in archived, streaming,
and live data, as well as to predict
future values and learn from examples
and observations at different levels of
abstraction. In the automotive sector,
modern statistical ML techniques are
used to develop new features, such as
mapping, intelligent control, autono-
mous driving, active safety, and prog-
nostics, and they have the potential to
touch every aspect of the vehicle
throughout its lifecycle, from design to
manufacturing to operation. A general
attribute of these algorithms is that
they are much too complex and non-
transparent for humans to understand.
People also cannot directly interfere
with their core mechanisms (for exam-
ple, by manually adjusting the weights
in a deep neural network).
AI Implications
The anticipated benefits of the next
wave of AI has encouraged politicians,
economists, and policy makers to pay
more attention to the technology. Rus-
sian President Vladimir Putin told stu-
dents, "Artificial intelligence is the
future, not only for Russia, but for all
humankind. It comes with colossal
opportunities but also threats that are
difficult to predict. Whoever becomes
the leader in this sphere will become the
ruler of the world" (https://www.rt.com/
news/401731-ai-rule-world-putin).
Like any other technology, AI is a
double-edged sword. Some authorities
consider superintelligence to be a sig-
nificant existential risk to humanity,
and others believe that it is beneficial.
Elon Musk, Tesla and SpaceX chief
executive officer, believes that competi-
tion for AI superiority at the national
level could lead to third world war, and
he initiated a movement (at the 2015
Puerto Rico AI conference) to ban
autonomous offensive weapons [7].
Stephen Hawking and Bill Gates
warned about the implications of AI
for humanity. In July 2017, the IEEE
and the IEEE-SA announced the IEEE
Global Initiative for Ethical Consider-
ations in Artificial Intelligence and
https://www.rt.com/news/401731-ai-rule-world-putin https://www.rt.com/news/401731-ai-rule-world-putin
IEEE Robotics & Automation Magazine - September 2019

Table of Contents for the Digital Edition of IEEE Robotics & Automation Magazine - September 2019
