A Letter from the Editor
AI Ethics: Science
Fiction Meets
Technological Reality
Daniel Zeng, University of Arizona and Chinese Academy of Sciences
Editor: Daniel Zeng,
University of Arizona and
Chinese Academy of Sciences,
zengdaniel@gmail.com
T
here’s no doubt that AI is becoming a permanent component of general
public consciousness: AI-enabled technologies are now commonplace in
mainstream applications.
We’d Like to Hear from You
Letters to the Editor:
Send letters, including a reference
to the article in question,
to bkirk@computer.org.
Letters will be edited
Growing Interest in AI Ethics
for clarity and length.
As the interactions between AI-enabled technologies and users (individuals, organizations, governments, and society at large) intensify and deepen, the interest
in the ethical aspects of AI and AI applications is rapidly growing. Major actions
are being taken by AI technology providers and governments to deal with such
issues. For instance, in early 2014, Google established an internal AI ethics board
to ensure the proper application of AI technologies. We suspect that it’s very likely
that other major players in the AI industry will soon set up similar mechanisms.
Multiple countries are deliberating legal guidelines governing the use of driverless cars—for instance, in the US, legislation has been passed in several states and
Articles:
If you’re interested in submitting
an article for publication,
see our author guidelines at
www.computer.org/
intelligent/author.htm.
2
On mobile devices, intelligent personal assistants such as Apple’s Siri, Microsoft’s Cortana, and Google’s Google Now are touted as a successful integration of a variety of AI technologies such as natural language processing,
knowledge discovery, intelligent interface, and recommender systems. A new
generation of cognitive systems, best represented by IBM’s famed Watson system, is opening up new venues for commonsense reasoning, knowledge engineering, and task automation. Autonomous driving, based on a number of AI
technologies such as robotics, planning, intelligent control, and multi-agent
systems, promises huge benefits in road safety. The expected reduction in traffic congestion and the likely major changes in human driving behavior and
mobility patterns, along with a potentially revolutionary impact on the auto
industry, have fascinated technologists and the general public alike. In military
contexts, autonomous logistical systems based on robotic vehicles of varying
kinds, and robotic combat capabilities, are already a reality or not far from it.
The past years have also seen exciting developments in AI methodology itself. Take the example of deep learning. In just a few years, deep learning
and the related end-to-end computing paradigm have blossomed into an active
field of study with major success in a variety of applications, rekindling the
interest in one of the old AI traditions concerning the development of humaninspired general problem-solving mechanisms.
It’s safe to say that a new era of AI is coming!
1541-1672/15/$31.00 © 2015 IEEE
Published by the IEEE Computer Society
IEEE INTELLIGENT SYSTEMS
IEEE
Washington, DC, to allow their use,
and in Germany, a committee consisting of researchers, practitioners from
the auto industry, and the government
has recently formed to draft a legal
framework to regulate such vehicles.
Research interest in AI ethics (broadly
construed) is growing, but ethical
studies of AI aren’t actually new. Before the establishment of AI as an academic discipline, philosophers, science
fiction writers, and futurists were already discussing the ethical ramifications of machine intelligence. Such
discussions are the precursors of AI
ethics studies. In today’s technological backdrop, old and new AI ethical
discussion threads alike, to a large extent, are no longer fantasy pastimes.
The legal, policy, and societal relevance and significance of AI ethics are
now almost palpable.
Major Discussion Threads
Here, I summarize the major discussion threads concerning AI ethics.
Needless to say, this summary isn’t
meant to be comprehensive. Rather,
I’m aiming at the ongoing hotspots of
meaningful discussions that might trigger further research, legislative actions,
or broader societal repercussions.
Discussion Thread 1: “Technological
Singularity and AI Doomsday
Scenarios”
From AI’s earliest days, philosophers,
technologists, futurists, and the g eneral
public have been fascinated and, in
some cases, frightened by the potential
of AI and its possibly drastic impact on
society and human civilization itself. A
stream of science fiction movies and TV
series dramatize various viewpoints,
ranging from AI doomsday scenarios to
AI/robots helping humanity prevail in
difficult situations or saving civilization
from the dark side of humanity itself.
Many important figures have jumped
into this discussion as well. Stephen
may/june 2015
Hawking is a leading voice cautioning
about the potential threats posed by AI,
famously saying, “The development of
full artificial intelligence could spell the
end of the human race.” Technologist
Elon Musk posits that the development
of AI is like “summoning the demon.”
Bill Gates has also joined them, suggesting that AI could be an existential
threat to humans.
At the other end of the spectrum,
several influential thinkers propose
the idea that AI could culminate in
human immortality and become a
phenomenon at the cosmological scale
redefining consciousness and matter.
Technological singularity has been
used to refer to what might happen as
artificial intelligence exceeds human
intellectual capacity and control.
The following statement, from Eric
Horvitz, director of Microsoft Research
Lab (and a member of this publication’s
advisory board), and his colleagues, represents a well-balanced and pragmatic
viewpoint: “AI doomsday scenarios belong more in the realm of science fiction
than science fact. However, we still have
a great deal of work to do to address
the concerns and risks afoot with our
growing reliance on AI systems.”
Discussion Thread 2: “Impact of
Automation on Economy
and Employment”
There’s no doubt that the application
of AI leads to increased productivity
in specific areas of the economy. Yet,
unintended adverse effects could occur. Take, for example, high-frequency
or algorithmic trading, which typically involves AI elements. Put in the
limelight by the US equity market
flash crash in May 2010, automated
trading practices have triggered a
lively debate. One school of thought is
that algorithmic trading has improved
the overall quality of markets through
lowered trading costs, more liquid
markets, and reduced discrepancies
www.computer.org/intelligent
IEEE Computer Society
Publications Office
10662 Los Vaqueros Circle, PO Box 3014
Los Alamitos, CA 90720-1314
Associate Manager, Editorial Services
Product Development
Brian Kirk
bkirk@computer.org
Editorial Management
Jenny Stout
Publications Coordinator
isystems@computer.org
Director, Products & Services
Evan Butterfield
Senior Manager, Editorial Services
Robin Baldwin
Digital Library Marketing Manager
Georgann Carter
Senior Business Development Manager
Sandra Brown
Senior Advertising Coordinator
Marian Anderson
manderson@computer.org
Submissions: For detailed instructions
and formatting, see the author guidelines
at www.computer.org/intelligent/author.
htm or log onto IEEE Intelligent Systems’
author center at Manuscript Central
(www.computer.org/mc/intelligent/
author.htm). Visit www.computer.org/
intelligent for editorial guidelines.
Editorial: Unless otherwise stated,
bylined articles, as well as product and
service descriptions, reflect the author’s
or firm’s opinion. Inclusion in IEEE
Intelligent Systems does not necessarily
constitute endorsement by the IEEE or the
IEEE Computer Society. All submissions
are subject to editing for style, clarity, and
length.
3
in prices across related markets. The
counterargument concentrates on a
mirage of negative impact: creating
something resembling a single point
of failure, benefiting the very few at
the expense of the very many, and
trading ahead of market orders to the
detriment of long-term investors.
At the macro level, the total return
on AI investment needs to be carefully weighed, especially when considering potential downsides such as
increased inequality and unemployment: increased automation could
push income distribution further toward the winners-take-all scenario.
However, although traditional economic agents—the providers of cheap
labor and “ordinary” capitalists—
might be increasingly squeezed by automation, a new group of people who
can innovate, design, and develop
new products, services, and business
models might emerge as winners in
the new knowledge economy.
Discussion Thread 3: “Legal
Ramifications and Accountability”
The current legal system, as well as
social norms and ethical standards
governing accountability and credit/
blame assignment, are made for humans and relatively low-tech technological underpinnings. AI-enabled
hardware and software systems, as
they’re embedded in the modern-day
societal fabric, are starting to challenge today’s legal and ethical systems.
For instance, in the current legal system, intent determines whether an act
is considered a crime. As autonomous,
intelligent, and adaptive AI entities
roam across both the physical and cyber worlds, how will we define crime
and intent? In particular, the era of
driverless cars will emerge in the notso-distant future, raising new questions in liability law. How can legislative bodies, jointly with technology
providers and consumers, and society
4
at large, design new legal frameworks
that encourage the realization of the
safety benefits in autonomous driving
technology? How much legal responsibility and protection should the driverless car manufacturers and designers enjoy? In the medical domain, who
will be responsible for accidents or errors made by robotic surgical systems
or automatic diagnostic systems?
Isaac Asimov’s classical work on
robots raised many interesting questions about the responsibilities of robots and how to interrogate the ones
that kill their human masters. In the
futuristic but entirely conceivable robot era, can a robot be put on a trial?
How much responsibility should the
robot’s designer assume? Can a robot
be called on to testify against itself?
Discussion Thread 4: “Privacy
Considerations and Human Rights”
Intelligent softbots and assistive robots
are already collecting and processing a
lot of information from their human
users in an array of task environments.
Privacy concerns abound in such cases
and have attracted a lot of attention.
For instance, Ryan Calo wrote a chapter in Robot Ethics: The Ethical and
Social Implications of Robotics (MIT
Press, 2011), discussing three aspects
of privacy: direct surveillance, increased
access, and social meaning.
The major implications of robotics
on human rights are becoming a reality. According to a Guardian report
published on 20 June 2014, the Skunk
Riot Control Copter, a drone armed
with plastic bullets and pepper spray
built by a South African company, has
been sold to an international mining
company interested in using it to suppress labor riots. “Shaking the Foundations: The Human Rights Implications
of Killer Robots,” a report released by
Human Rights Watch and Harvard
Law School’s International Human
Rights Clinic on 12 May 2014, finds
www.computer.org/intelligent
fully autonomous weaponry could
pose major challenges to fundamental human rights. Such concerns have
also been discussed in high-profile international events—for instance, the
first multilateral talks on killer robots
opened at the UN in May 2014.
Discussion Thread 5:
“Human-AI Relations”
Although humanoid robots are still a
work in progress, it’s not inconceivable that such robots or AI machines
assuming other outward forms will
interact with humans holistically in
the future, including at the emotional
level. As such interaction intensifies
and grows, interesting ethical considerations concerning human-AI relations will arise. Can robots really
understand our feelings? Do we have
feelings for robots?
In recent years, a growing number
of robots have been developed to care
for or accompany children, the elderly,
and the disabled. Without a doubt,
these robots will provide a lot of benefits. However, it’s entirely possible that
their users could develop various forms
of attachments to these robots. Such
attachments could lead to interesting
and challenging relational and ethical
issues, warranting further research.
Discussion Thread 6: “Robot Rights”
Humans have human rights. Animals have animal rights. Should robots have robot rights? Should they
be treated as conscious beings? Does
the First Amendment protect robots’
speech? Do we need laws to protect
robots from being abused?
It’s still too early to formulate specific answers to these “robot rights”
questions given the current state of
the art of robotics. Nonetheless, these
questions are already drawing significant attention from the general public.
In February 2015, heated discussions
erupted in both mainstream and social
IEEE INTELLIGENT SYSTEMS
The MIT Press
media about a video showing Boston Dynamics employees
kicking a robot dog to show its robustness. A widely held
opinion is that although this dog isn’t a real, living thing,
kicking it is inappropriate and can be construed as a violent
behavior. A prevailing thought is that although we don’t necessarily need to care about robot rights, abusing robots (just
as abusing animals) very likely will make people more abusive toward other people. As such, protecting robot rights
indirectly protects human rights. The emerging field of roboethics, short for robotics ethics, is concerned with these and
other issues, such as how “humans design, construct, use and
treat robots and other artificially intelligent beings.”
D
iscussions about AI ethics are fascinating, thought-provoking, forward-looking, and in no small part entertaining,
but one thing is clear: AI ethics is no longer a purely philosophical topic. Nor does it still fall under the realm of creative
thinking by science fiction writers and futurists. Because of the
increasingly deeper and wider adoption of AI technologies, AI
ethical issues are becoming much more real, with practical relevance and significance. This emerging field calls for rigorous
research and collaborative efforts from disciplines as disparate
as philosophy, law, anthropology, economics, politics, computer security, and, of course, different branches of AI itself.
Most people with more than a cursory interest in AI
are familiar with Asimov’s famed three laws of robotics.
Against the backdrop of a newly kindled interest in AI ethics, several researchers in Europe argued a few years ago
that an extended set of ethical rules are needed for building
robots. Five laws were put forth:
• “Robots should not be designed solely or primarily to kill
or harm humans.”
• “Humans, not robots, are responsible agents. Robots are
tools designed to achieve human goals.”
• “Robots should be designed in ways that assure their
safety and security.”
• “Robots are artifacts; they should not be designed to exploit vulnerable users by evoking an emotional response
or dependency. It should always be possible to tell a robot
from a human.”
• “It should always be possible to find out who is legally responsible for a robot.”
ANIGRAFS
Explorations of a Collaborative
Cognitive Architecture
Whitman A. Richards
INTELLIGENCE EMERGING
Adaptivity and Search in
Evolving Neural Systems
Keith L. Downing
An innovative proposal for
understanding how mental
organisms make decisions and
control behavior.
An investigation of intelligence
as an emergent phenomenon,
integrating the perspectives of
evolutionary biology, neuroscience, and artificial intelligence.
128 pp., 7 color illus., 49 b&w illus., $25 paper
Of course, we can always argue whether these laws or
certain components of them make sense or not. But as to
direction, such potentially actionable guidelines and shared
understandings might be what the AI community needs
to further push the technological envelop while remaining
mindful about our humanistic and societal bases.
may/june 2015
www.computer.org/intelligent
508 pp., 196 illus., $50 cloth
mitpress.mit.edu
5
Purchase answer to see full
attachment