From the
PRESIDENT
Robert A. Zaring, MD, MMM
GLMS President | [email protected]
AI: THE FUTURE IS HERE
W
hat do Stephen Hawking,
Elon Musk, Jeff Bezos, Mark
Zuckerberg and Sundar Pichai
have in common? The obvious
answer is they are all titans
of science and technology. However, they are
all also making comments lately regarding
artificial intelligence (AI). Pichai, Zuckerberg
and Bezos have praised AI’s potential to
transform the world in positive ways. In
contrast, Hawking and Musk have offered
warnings. Hawking said, “The development
of full artificial intelligence could spell the
end of the human race.” Musk has shown his
concerns by stating that part of the reason for
his company, SpaceX, is so that humans can
colonize other planets to escape Earth - when
AI takes over. These warnings seem at first to
be so farfetched and Hollywood-inspired that
it’s difficult to take them seriously.
However, the advancements in AI made
over the last few years have been dramatic,
and although we are not interacting with Hal
as in “2001: A Space Odyssey,” we do regularly
interact with Siri and Alexa. Numerous ethical
questions come to mind when we think of the
fact that we may be on the way to reaching
the point when computer intelligence equals
human intelligence. That may seem a long way
off, but noted futurist, computer s cientist, and
Google’s director of engineering, Ray Kurzweil,
has predicted that to be in the year 2029,
and a merger of machine and man in 2045.
Interestingly, Elon Musk has recently started
talking about a future “hybrid man” as the only
way mankind can deal with AI.
AI will also affect local economies. Kai-
Fu Lee, the founder of venture capital firm
Sinovation Ventures and a leading voice of
technology in China, has said AI will replace
50 percent of all jobs in five years. Considering
these enormous ethical dilemmas, it is not
surprising that Google, Facebook, Amazon,
IBM and Microsoft have come together to
create the Partnership on Artificial Intelligence
dedicated to looking at the ethics of AI.
With such huge potential changes in
everyday life, I wonder what the medicine
of tomorrow will look like and how AI will
change health care. To understand what
those changes will look like, it is important
to understand some basic concepts of AI. The
term “AI” arose from computer scientists at the
Dartmouth Conferences in 1956. At that time,
they envisioned a computer that thought like a
human. Through the years, most of the dreams
of AI were exactly that: dreams. But, the vision
was there.
What came next was machine learning,
which is the parsing of data by algorithms,
learning from the data and then making
predictions. This was labor intensive and still
did not allow for “Narrow AI” or technology to
do a specific task as well or better than humans.
What changed that was deep learning, which
has allowed AI to explode since 2015. With
the use of parallel processing, neural networks,
graphic processing units and machines
learning, a dazzling amount of data could be
analyzed and train each node in the neural
network. That is exactly what Andrew Ng of
Google Brain did, allowing the computer to
recognize cats by viewing millions of Youtube
videos. It is also how Google DeepMind’s
AlphaGo program beat South Korean Master
Lee Se-dol in the board game Go in 2016.
The AlphaGo program learned the game and
trained itself by tuning its neural network by
playing itself over and over.
So, you are probably wondering what this
has to do with medicine. This brings me to a
discussion of two of the more noteworthy uses
of AI in health care currently, namely Google’s
Deepmind Health project and IBM’s Watson.
Google’s Deepmind Health has partnered with
the UK’s NHS and in one project teamed up
with Moorfields Eye Hospital NHS Foundation
Trust to examine millions of eye scans. This will
allow the AI to learn and be able to diagnose
two main eye conditions, wet age-related
macular degeneration and diabetic retinopathy.
It is believed that Google’s Deepmind will be
able to help physicians dramatically in precisely
diagnosing serious eye conditions, at a greater
rate. While one of Google’s Deepmind projects
is working on eye disease, IBM’s Watson is
tackling cancer. Watson used case histories and
medical literature to be able to recommend
treatment plans for individual patients. In
2016, experts at the University of North
Carolina School of Medicine tested Watson by
having it analyze 1,000 cancer diagnoses. In
99 percent of the cases, Watson recommended
treatment plans that matched suggestions from
oncologists. Watson also identified treatment
options doctors missed in 30 percent of the
cases.
As good as AI seems to be with logic and
data, can it really replace the physician? Those
of us who practice know nothing can replace
human compassion and the patient-physician
relationship. So, the answer appears to be no.
Can AI be a significant aid for physicians in the
battle against disease and suffering? The answer
is most certainly yes. However, physicians
must be engaged in the discussions of the use
of AI and its ethical considerations. It can
become too easy for technology to surpass our
ability to regulate it or consider all the ethical
implications and protect against the negative.
It is admirable that giants such as Facebook or
Amazon are looking at the ethical concerns of
AI, but both have large profit motives for this
technology. AI will almost certainly change our
practices, and you wonder if someday we will
be concerned that a patient is being infected by
malware.
Wherever AI takes medicine, physicians
must lead in the discussions of uses, integration
into practice and ethical constraints. There is
more to treating humans than data and logic,
and there is nothing like human compassion
for healing someone. We must not let ourselves
or others ever forget that.
Dr. Zaring is an anatomic and clinical pathologist
with Louisville Pathology Associates and practices
at Jewish Hospital.
SEPTEMBER 2017
5