Letting a computer make automated decisions in ethical problems is

advertisement
IT, ethics and organizations 5p
Assignment 2 for seminar 2
Johan Bergman, Hanna Chamoun, Staffan Gustafsson, Ismail Mechbal
Letting a computer make automated decisions in ethical problems is technically possible to
some extent, but doesn’t guarantee that the outcome is reliable. The inputs required for the
computer to make any decision should be based on the process of how the problem owner
makes decisions. Human decision making relies on the problem owner’s personal history,
knowledge, emotions, political views etc. So if a computer should be able to automatically
perform the decision making of a moral problem, the computer must be able to somehow
know everything about the person owning the problem. Only then, the computer has all the
knowledge needed to draw a "good" conclusion.
A computer is better at solving general problems than humans are since the computer is much
faster when it comes to calculations. The computer is also not limited when it comes to
information storing and searching; thus, the computer will be able to remember and compare
all information needed to solve a problem.
For moral problems the computer therefore needs to be able to receive and calculate all the
information the person has experienced in all her life. This must be done in such a way that
the computer can measure the person's feelings about the information; there need to be some
sort of tagging and values put to the data.
When confronted with an ethical problem the computer will then have a very big database of
information, feelings, connections and weights of importance. If it is possible to do as Bran
says, to give each possible solution a weight and a direction, then one can insert the data with
these weights so that the computer can calculate which option that is the best one for certain
problems. The computer also needs to be able to combine all possible variables and put the
right variables together to create an outcome of something that is very fluctuating, and in the
end dependant on feelings and values which both are very far from as black and white as
computer calculations are. So even if you could use Bran’s ideas to compare possible
solutions the computer will only know how to do so from the data that is available without
reflecting on current events and feelings that we as humans add to the solution of an ethical
problem.
In our opinion, a moral computer without all the knowledge from the problem owner and the
world in which the problem is situated is not capable of drawing any good conclusions when
it comes to moral decision making. Maner's algorithm assumes that problems are "black and
white". He even himself have seventeen different aspects that different people may see as
problems with his algorithm. How can the algorithm know what weight of importance a
possible solution should get? This is something that requires knowledge from real life which
changes from day to day. Also, if Maner's computer would require the user to enter weights
and other information, we think it would literally take ages to enter all data necessary for the
calculation.
The risks with using a computer are that the computer cannot think of certain aspects that the
human can do very quickly and easily, which can result in an unreliable outcome. Another
risk is that the outcome of a computer could be overruled by your own thinking, because
people tend to get lazy when a solution is already presented. This leads to us trusting the
computer without critically examining the steps taken to get the result.
The only benefit is that the computer could help you, by giving an overview of problems
without any solutions. The fact that a computer is faster and can store more data than a human
IT, ethics and organizations 5p
Assignment 2 for seminar 2
Johan Bergman, Hanna Chamoun, Staffan Gustafsson, Ismail Mechbal
while its lack of being able to critically consider the valuable aspects of moral decision
making makes it a very good tool for helping us see the whole problem so that we can solve it
and take the critical decisions, the computer should not take them for us.
We see the stages that Manner presents as a good list of things that needs to be done to be
able to understand the problem. It is very similar to the general aspect of HCI when you
collect data, try to understand the problems and situations and lastly create a system for
supporting the work. To be able to critically understand and create a solution you need to
know the background and all the aspects of the problem. You yourself are seldom the only
one inflicted by the problem and it’s not often that you are the only one who should have
something to say. You need to know as much as possible about the situation. This is where
Manner’s list of stages comes in. It is a good way of helping you get the data needed, and help
you not to miss important aspects.
The stages are good in theory, but, to keep track of all possible outcomes and all aspects and
then weigh them against each other, and against all other parties, and within each party and so
on seem like a never ending story. Computers can help us choose the right idea by presenting
a lot of data but the data needs to be unbiased and as complete as needed. All stages up to the
decision making can be done with the help of computers in some sense. But when it comes to
the weighing and deletion of aspects the problem owner needs to be the one that takes
responsibility and action to choose the best ideas.
The computers should only be used as storage facilities since we as humans only can keep a
certain amount of information in working memory at once and more important; moral
problems need to be solved by humans.
IT, ethics and organizations 5p
Assignment 2 for seminar 2
Johan Bergman, Hanna Chamoun, Staffan Gustafsson, Ismail Mechbal
Using an automatic
moral computer
Decision maker
Primary stakeholders
Third party
stakeholders
Society
Using tools for
ethical decision
making
Decision maker role is Help finding decision,
buried, no importance
less hard work to
toward human
find appropriate
decision, more danger
decision, greater
for wrong decision
chance of getting the
“best” decision
Solving the problem
without help from
tools
All emphasis on the
decision maker and
his abilities and
knowledge, chance of
getting the usual
decision with no
critical thinking
Danger of wrong
critical decision, no
importance to human
decision, their
concerns are probably
not met
Help finding good
Get the same solution
decisions in short
as usual
time, their concerns
are probably thought
of, a good overview
of why the decision is
taken
Danger of wrong
critical decision, no
importance to human
decision, their
concerns are probably
not met
Help finding good
Get the same solution
decisions in short
as usual
time, their concerns
are probably thought
of, a good overview
of why the decision is
taken
Danger of wrong
critical decision, no
importance to human
decision, their
concerns are probably
not met
Help finding good
Get the same solution
decisions in short
as usual
time, their concerns
are probably thought
of, a good overview
of why the decision is
taken
Download