Thousands of third-party contract workers have been poring over private conversations recorded by Alexa-enabled devices (for the purpose of improving Alexa’s speech recognition systems). Is this ethical? How can AI software be trained while respecting privacy?
Posed by a journalist.
To comment on this dilemma, leave a response. For anonymity, omit your email address and website, and use a screen name.
In September 1935, the Nazi Party asked Bernhard Lösener to draft a large portion of the infamous Nuremberg Laws, which deprived Jews of citizenship rights and eventually led to the Holocaust. Lösener was clever enough to convince party officials that a person should be considered Jewish only if he/she had at least three, rather than two, Jewish grandparents. Lösener later argued that by collaborating with the Nazi regime, he was able to save thousands from the Holocaust by reclassifying them as non-Jews. Indeed, by one estimate, he saved as many as 100,000 lives.* If Lösener had not collaborated, someone else would have written the Nuremberg Laws, and they would have been even worse. Does this justify his actions?
*Robert Nelson, Revolution and Genocide, University of Chicago Press (1992) p 324.
Thanks to David Goldman, founder of FASPE, for suggesting this dilemma.
To comment on this dilemma, leave a response. For anonymity, omit your email address and website, and use a screen name.
I pass the Tree of Life synagogue in Pittsburgh every day, on my way to and from work. Yesterday, a gun-wielding assailant entered the synagogue and murdered 11 participants in a Shabbat service, shattering the peace of the richly multiethnic community of Squirrel Hill. The killer had posted messages online expressing his hatred of Jews and immigrants. Here was the full horror of hate, in my own neighborhood.
There is a temptation to dismiss the seemingly endless hate crimes we have endured in recent times as the work of isolated psychopaths. But we all know this is a mistake, because we all know about the rising tide of hatred that underlies these acts. What can we do about it?
To comment on this dilemma, leave a response. For anonymity, omit your email address and website, and use a screen name.
Virtues (in the Aristotelian tradition) are qualities that humans alone bring to the world — and therefore explain, in some sense, why we are here.
Classical virtues include courage, honor, loyalty (to other people), applied intelligence, aesthetic sensibility, and sophrosyne (an ability to find the right balance when virtues conflict).
You act inconsistently with your purpose as a human being if you sacrifice one of the virtues. Virtue ethics says that you should compromise a virtue only when necessary to preserve another virtue.
Integrity is (literally) wholeness. You lose integrity when your actions are not virtuous, because they conflict with who you are as a human being.
For a fuller explanation, watch the video Rational choice II (or read the transcript). It is part of an online ethics tutorial that consists of the following 30-minute sessions:
The autonomy principle says that you shouldn’t interfere with freely chosen, ethical actions of other people without informed or implied consent.
This means, for example, that you shouldn’t kill others. Serious injury, enslavement, and coercion violate autonomy when they interfere with freely chosen, ethical actions. Respect for autonomy can be more precisely defined by the following principles:
Principle of joint autonomy. You should not do anything that you are rationally constrained to believe is jointly incompatible with the freely chosen, ethical actions of a group of one or more other persons. Throwing a bomb into a crowded street violates autonomy, even if there is no one person you know will be harmed by the bomb, because you know that someone on the street will be harmed.
Principle of interference. Interfering with unethical behavior does not violate autonomy, because unethical behavior is not freely chosen action. It has no coherent rationale and therefore cannot be distinguished from behavior that is not deliberately chosen. For example, you can grab a mugger to prevent him from attacking a victim. However, the interference must not go so far as to prevent ethical action, unless there is implied consent (see next point). For example, you can’t lock someone in a closet to prevent him or her from cheating on income taxes, because this interferes with many ethical actions. Also, the principle refers to interference by an individual, not necessarily by the state. Further argument is necessary to determine when and whether incarceration is justified, for example.
Principle of implied consent. You can interfere with a person’s ethical action, without violating autonomy, when that person implicitly consents to a policy that is consistent with the interference. You can tackle a distracted pedestrian who is about to walk into oncoming traffic, because the pedestrian already has a policy of not walking into traffic and therefore implicitly consents. You can incapacitate a mugger, when necessary, to protect yourself from attack, even though this interferes with ethical as well as unethical actions, because the mugger already has a policy of self-defense and therefore gives implied consent.
Principle of informed consent. You can expose a person or group of people to harm, without violating joint autonomy, when there is informed consent. You can administer a vaccine to 1000 people even if you know that it will make someone ill, if all 1000 give informed consent.
The autonomy principle doesn’t require you to give people anything they want; it only requires you not to interfere with free and ethical choices. Someone may want to use your car, but refusing to let them use it is not a violation of autonomy, because they can’t choose to use your car. They can only choose to ask permission use it. True, they can choose to use your car if it is unlocked and you leave your keys in it, but that choice would be unethical because it is illegal. Preventing them from getting into your car to drive it away would not be a violation of autonomy.
For additional explanation, watch the video Rational choice II (or read the transcript). It is part of an online ethics tutorial that consists of the following 30-minute sessions:
The utilitarian principle says your action is ethical only if you can rationally believe that it maximizes the total expected utility of everyone it affects. It must create at least as much utility as any other option that satisfies the generalization principle and the autonomy principle.
Utility can be understood as happiness, or any other condition(s) you regard as intrinsically valuable. It is the ultimate end (or ends) toward which all of your actions are means.
Total utility is the algebraic sum of everyone’s utility, where negative consequences are regarded as having negative utility.
When the consequences of an action are uncertain but probabilities can be estimated, expected utility is the weighted average of utilities in all possible outcomes, where the weights are the probabilities of the outcomes.
An act that is consistent with the utilitarian principle can be said to pass the utilitarian test.
For a fuller explanation, watch the video Rational choice II (or read the transcript). It is part of an online ethics tutorial that consists of the following 30-minute sessions:
The generalization principle requires that the reasons for your action be consistent with the assumption that everyone with the same reasons acts the same way.
The reasons for your action are not psychological causes or motivations, but the reasons that you consciously take to justify taking the action. They must be sufficient and necessary, meaning that you would perform the action whenever all the reason apply, and you would not perform it if any one reason failed to apply.
An act that satisfies the generalization principle is said to be generalizable or to pass the generalization test.
Your act passes the generalization test only if you can rationally believe that the action would still achieve its purpose if everyone with the same reasons acted the same way.
For example, if you tell a lie simply because it is convenient to deceive someone, your reasons are: your lie will deceive someone, and the deception is convenient. This is not generalizable, because you cannot rationally believe that your lie would deceive anyone if people always lied when deception is convenient. People would assume you are lying like everybody else.
For a fuller explanation, watch the video Rational choice I (or read the transcript). It is part of an online ethics tutorial that consists of the following 30-minute sessions:
My cousin has been helping a friend with depression for 6 months. However, my cousin started school recently and has a very busy schedule of 6 courses and a job. She is reluctant to leave her friend alone with her suicidal thoughts, because she has no one else to turn to. Yet she feels that she would be wronging herself to neglect her self-development. Should she sacrifice her time for her friend or focus on herself, given that she can’t do both?
Contributed by anonymous.
To comment on this dilemma, leave a response. For anonymity, omit your email address and website, and use a screen name.
A few years ago, I registered for a driver’s license in my home country. I signed up for the compulsory classes and exams, which were expensive. After classes were completed, it was time for the driving test. Everything went smoothly, and I completed the test. When I went to retrieve my license, the official behind the counter told me that the results of the test would be released by the end of the day. However, if I wanted to pass, I had to give him a “tip” of a significant amount. He mentioned that everyone else gave him a “tip,” as anyone who did not would fail the driving test automatically. Should I pay the bribe?
Contributed by anonymous.
To comment on this dilemma, leave a response. For anonymity, omit your email address and website, and use a screen name.
Ethical Decisions 