June 12, 2012 16:42 PSP Book - 9in x 6in 14-Junichi-Takeno-c14
Appendix B
On the Safety and Ethics of Robots
Many people are concerned about the hazards of robots having
“consciousness” or a “mind.” In the movie The Terminator,a
computer system called Skynet becomes aware of itself and plans
and executes the extermination of humans considered to be
“imperfect biosystems.”
Machine systems never die and their capabilities exceed those of
humans in all aspects. Modern machine systems are more intelligent
than humans at least in terms of their memory capabilities and
their speed of computation. Machine systems, by their nature,
have unsafe aspects. It is possible that robots could lose self-
control and run amok. This actually happened in Japan where
a playback robot in a plant ran out of control and squashed a
worker. Although this kind of serious accident should never happen,
minor similar accidents occur almost daily. One typical example is
runaway automatic transmission cars. Most of these incidents are
human errors. Runaway robots can be prevented to some extent
by controlling them according to some legal safety standards. For
example, high-level international safety standards are in place for
cars and elevators.
Conscious robots proactively collect information, have their own
values, and behave according to their will. Robots may have values
that are not favorable for humans and could interfere with people’s
daily life. Robots may commit a murder intentionally in extreme
cases. To eliminate the possibility of these negative aspects, we
June 12, 2012 16:42 PSP Book - 9in x 6in 14-Junichi-Takeno-c14
248 Appendix B: On the Safety and Ethics of Robots
should construct a consciousness system from the ground up that
has robots abide by the three principles of robots as set forth by
Isaac Asimov and shares the values with, and behaves under control
of, humans. Robots so constructed would suppress any behaviors
that would embarrass humans and thus would not bring about any
serious problems.
If a robot should appear that is educated by humans to commit
a crime, it must be detained as a human would be and should be re-
educated. This is not plausible, however, because if such a robot were
developed, the developer’s life would be at risk. If the suspect robot
were difficult to “detain,” physical destruction would be an option.
Even when robots are destroyed, no “human rights” issues, such as
in the case of human clones, would be involved. If any problem oc-
curred at all, it would be a “problem of the mind” on the human side.
Human clones are obviously humans as a biosystem, and human
rights issues would occur if they are defamed in any way. I believe
that no “human rights” issues would occur with robots because they
are built as a machine system. Instead, a new issue of “robot rights”
may arise. Robots could argue that they have the right to exist if they
face the risk of destruction. The destruction of a robot is generally
believed to be not “destruction without reproducibility.” Thus, it
is possible, at least in principle, to “copy a robot” by extracting
all of the data from the brain of the robot before destruction, and
downloading it to another robot that is physically identical.
Let’s go one step further with this idea. I mentioned the problem
of copying a robot in the previous paragraph. Note that I meant
“copying” the self, and not “reviving.” The original self of the
destructed robot would be lost eternally and the new robot with the
copied self will lead a new life. This is the logical consequence as
derived from my theory.
It is utterly difficult for a third-party observer to differentiate
between a copied and a revived robot. There is, however, a big
difference for the robots. The copied robot is a new one, i.e., it is
different from the original destroyed one.
Let’s take a simple example. Here is a sheet of a manuscript. You
copy it on a copier. Now you have two separate sheets of manuscript
with identical contents. I would like to indicate that there is no
correlation between these two sheets of paper.
June 12, 2012 16:42 PSP Book - 9in x 6in 14-Junichi-Takeno-c14
Appendix B: On the Safety and Ethics of Robots 249
In the near future, discussions on the eternal life of robots
will emerge as a problem that cannot be solved by present-day
science, just as the eternal life of humans is today. I believe that it is
unavoidable for researchers of conscious robots to face the problem
of robot rights in the near future. It is apparent that robot rights will
be a serious problem in the future.
My conclusion is that conscious robots are controllable by
humans because their consciousness system is, in every detail,
understood by humans. In addition, the development of conscious
robots is expected to bear significant fruit for humankind — the
understanding of the mechanism of human consciousness. Nuclear
energy and biotechnology are dreadful technologies in the sense
that once human control is lost even slightly, their destructive forces
could endanger the whole world. Different from those technologies,
conscious robots should be developed as a useful means for humans,
just like cars, airplanes, and other convenient machine systems.
This page intentionally left blankThis page intentionally left blank
..................Content has been hidden....................
You can't read the all page of ebook, please click here login for view all page.