IEEE Power & Energy Magazine - May/June 2022 - 90

in my view (continued from p. 92)
Into the Future
If the historical cycle holds, the power
industry is set for another major disruption
within the next 40 years as the
push to decarbonization ramps up. I
would argue that three key technologies
will enable the paradigm shift in
the power industry toward 100% renewable
power systems.
AI
AI as we know it today was put on the
map by Alan Turing in his seminal article
in the 1950s, in which he argued that
machines could think and reason like a
human. The first proof of concept for AI
was showcased when Logic Theorist, a
computer program designed to solve
problems, was presented at a 1956 conference
cohosted by John McCarthy, an
American computer and cognitive scientist.
Research programs over the following
years gradually established AI
in general and artificial neural networks
in particular as a scientific discipline.
The first functional, and arguably
deep, neural network was published in
1965 by the Ukrainian mathematician
Alexey Ivakhnenko. Advancements in
electronics led to better computation
power and memory in
the 1980s, paving the
way for the development
of practical neural
networks. Powered
by the backpropagation
algorithm, neural
networks reached
a milestone in 1986
by predicting the next
word in a sequence.
By the end of the 20th
century, several practical
research areas were
established that focused
on solving practical
problems and not
necessarily building a
machine " that could
reason. " This is how new terminology,
such as machine learning, deep learning,
data mining, and data science was
established in the day-to-day vocabulary.
Despite their sometimes-distinct
differences, the terms are often used
interchangeably.
However, a few developments in the
If the historical
cycle holds, the
power industry
is set for another
major disruption
within the
next 40 years
as the push to
decarbonization
ramps up.
early years of the new century led to
the deep learning revolution
as we know it
today. Geoffrey Hinton
and his team invented
methods to train multilayer
deep networks,
one layer at a time, cutting
training times significantly.
Moore's law
of exponential growth
in computational performance
and memory
began to economically
deliver enough computing
power and memory
to enable the spread of
AI applications. Collected
and consumed
data became truly big
and passed the zettabyte threshold by
2010. Andrew Ng figured out how to
use GPUs to speed up model training
by orders of 100. These advancements
led to the deep learning " big bang "
in 2011 when, for the first time, a fast
GPU implementation of convolutional
calendar (continued from p. 89)
July 2023
IEEE PES General Meeting (GM
2023), 16-20 July, Orlando, Florida,
United States, contact Roseanne Jones,
roseanne.jones@ieee.org
For more information on additional technical
committee meetings, webinars,
and events, please visit our IEEE PES
calendar: https://www.ieee-pes.org/
meetings-and-conferences/conference
-calendar.
p&e
letters to the editor (continued from p. 88)
interconnectivity with resources across
the entire United States can offer significant
benefits to make up for generation
losses in one or more regions.
The other options could include
continuing to use gas plants for capacity
needs (for example, targeting 90
90
ieee power & energy magazine
instead of 100% clean electricity) and/
or using renewably generated fuels or
other long-duration storage resources.
There are many challenges to overcome
to reach clean, affordable, and reliable
electricity, but there are also many potential
solutions. The biggest risk to a
future grid powered by wind and solar
may not be extreme weather, but simply
a calm, cloudy day.
-Debra Lew
Nov./Dec. 2021 Guest Editor
p&e
may/june 2022
https://www.ieee-pes.org/
IEEE Power & Energy Magazine - May/June 2022

Table of Contents for the Digital Edition of IEEE Power & Energy Magazine - May/June 2022
