(* NEW *) Here is another excellent Asimov page by Roland Saekow
- All you ever wanted to know about Isaac
Asimov. Find it on the
Isaac Asimov Home Page.
Isaac Asimov 1920 - 1992
Some of Isaac Asimov's most
Remarks on Asimov's Three Laws of Robotics
Isaac Asimov, who, in my opinion, is the greatest writer of all time, published his
three laws in a short story called "Runaround" which was published by Street and
Smith Publications, Inc. in 1942. The three laws were stated as follows:
- A robot may not injure a human being, or, through inaction, allow a human being to come
- A robot must obey the orders given it by human beings except where such orders would
conflict with the First Law.
- A robot must protect its own existence as long as such protection does not conflict with
the First or Second Law.
Several readers have asked about Asimov's Zeroth law. Finally, I have found time to
include it. The following paragraph was quoted from Rodger Clark's page on
Asimov detected as early as 1950, a need to extend the first law, which protected
individual humans, so that it would protect humanity as a whole. Thus, his calculating
machines "have the good of humanity at heart through the overwhelming force of the
First Law of Robotics" (emphasis added). In 1985 he developed this idea further by
postulating a "zeroth" law that placed humanity's interests above those of any
individual while retaining a high value on individual human life.
Zeroth law: A robot may not injure humanity, or, through inaction, allow humanity to
come to harm.
Notwithstanding my great admiration for Asimov, I do not agree with his laws (nor do
many other roboticists or sci-fi writers - such as Robert J. Sawyer).
Consider the first law. The first law precludes several very important
"careers" for which the android is well suited - namely: soldier, policeman, and
It is clear that the military will be quite interested in having android soldiers in
order to spare the lives of our human soldiers - and very likely to reduce the numbers of
human soldiers needed by the military thus reducing the national defense budget.
Androids would also be well suited as policemen. My major complaint against policemen
is that they shoot first and ask questions later. Hardly a year goes by in which we don't
hear of some cop killing a child who threatened him with a toy gun. I find that behavior
unacceptable. We also very often hear of policemen who kill people who are unarmed simply
because the cop "thought he was reaching for a gun". I find that behavior
unacceptable too. Rarely are the cops prosecuted and, who knows if the cop simply wanted
to kill that person for some reason - such as his skin color. Since an android is
not alive, he can't be killed and therefore he NEED NOT shoot first.
Androids would also make good security guards. Again, they need not shoot first. In
addition, they don't need sleep and won't get tired or hungry. They will always attend to
their duties and not "goof off." Androids don't need to eat so you will never
find them sitting in a doughnut shop. They can't be distracted by common events which
might affect humans - such as sporting events or pretty girls.
Consider now, the second law. While I generally agree with it, I believe it needs to be
reworded. Perhaps it could be written as: A robot must obey the orders given to it by its
owner or other human beings or androids designated by the owner. In its original form we
would encounter such bizarre situations as humans hijacking androids from other projects
and putting them on their own projects. You would be unable to count on your android
finishing his assigned tasks because other humans would order them to do something else.
Another consideration for which I currently have no solution, is whether the android
should follow orders which are immoral or unethical or illegal. Of course, I don't want
androids to engage in illegal activities, but it will be very difficult to prevent that
from happening. Violence (as in the first law) should be pretty easy to avoid, but immoral
or unethical activities will be much harder to prevent. Perhaps these questions are best
left to be determined by a special conference. I once had a friend who thought that as
long as something was legal, it was acceptable behavior. I do not subscribe to that view.
And finally, let's consider the third law. This is clearly the weakest of the three
laws. We humans of course protect our existence as the number one priority. But, for the
vast majority of us, this simply amounts to avoiding any of the many ways you can injure
yourself or be injured by others or get into traffic accidents. Common sense dictates that
the androids would do the same. Therefore, the third law is not really needed, because the
androids will never make the careless mistakes humans make and therefore their existence
will never be in danger. However, the case of emergency behavior must be planned for.
These are situations where the property or family of the owner are in danger. These
situations will be covered in the "startup procedure" of every android. The
owner will simply define a list of people and/or property for which the android will be
responsible. The android will then attempt to save those people and/or property in an
There will no doubt be many people who will hate androids because the androids will
displace them from their jobs. Some will hate them enough to try to destroy them. What
then will an android do when a human being sets out to destroy him? It appears that
according to the first law, the android will not be able to defend himself because he may
"injure" the human who is trying to destroy him. If I were an android owner, I
would not want anti-android people destroying him. First, the android is expensive and
second he/she will be doing useful work for me. I may have become emotionally attached to
the android too. It would be like a member of the family. In short, I believe the android
should be allowed to defend itself against anyone other than the owner and any other
humans designated by the owner. On the other hand, I do not think it would be a good idea
for the android to become a "hit man" or to otherwise attack humans on the
orders of its owner or his designated operators - unless of course, the android is a
soldier or the other human is attacking the owner or one of his designated
operators. It may be difficult, however, for the android to determine if it is being
asked to attack someone (see first example given below).
Finally, let me leave the reader with a few thought experiments in which the original
three laws would have difficulty.
- Suppose that you have packed a briefcase with explosives and a remote controlled
detonator. Then you call your android and give him the briefcase and a sealed letter. You
then direct the android to walk across the street carrying the briefcase and hand the
letter to Mr. Jones who is standing over there. The android follows your orders and walks
across the street. As he hands the letter to Mr. Jones, you explode the bomb by remote
- An android comes upon two humans fighting. What does he do? By interfering, he may
injure one of them, but by not interfering, they may injure each other.
- Several androids are working around a downtown building when the fire alarm sounds.
There are people in the building who may be burned up. Do the androids all run into the
building to try to save the humans - and end up burned up in the fire? How will the
androids know if they have rescued all the occupants? Do they extract dead bodies
before they get burned up?
- Several androids are walking across a bridge when they see a human jump off the bridge
into the water below. Do all the androids jump off the bridge to try to rescue the human?
What if the androids themselves can't swim or are too heavy to float?
I would be interested in your comments. Please
send email to: