L10: Ethics

Question 1

WHY BIG DATA ANALYTICS CAN BE
ETHICALLY PROBLEMATIC

Answer 1

BD analytics is powerful
* (Otherwise we would not care, would we?)
* Powerful tools always open possibilities for
misuse
* The role of ethics in data analytics is in
general often underestimated (because it
has been considered

Question 2

Responsibility when using BDA

Answer 2

• Traditional decision maker (Physician uses traditional means and mistreats patient)
• Decision makers using ML (Physician uses machines outcome and mistreats patients)
• ML Algorithm (Self-driving car crashes)

–> Who to blame? responsibility, legal

Question 3

ETHICAL CONCERNS SERVE NOT
ONLY A SELF-PURPOSE

Answer 3

Legal consequences
* Discrimination based on gender, race etc.—unintended or not—is illegal in most
jurisdictions
*Conformity with legislation, e.g., Data Protection

Regulation Impact on employees
* Homogenization of workforce (when hiring based on
ML decisions)
*Frustration

Impact on
customers
* E.g. Racist
outcome at

Google Photo
*Spillover to unaffected

customers Impact on business
partners
* Loss of trust, dissolving of
personal relationships

Question 4

EXAMPLE FOR THE POWER OF BIG DATA
ANALYTICS

Answer 4

Facebook Likes can
predict:
* race
* gender
* Men’s sexual orientation
* Women’s sexual
orientation

2013
95%
accuracy
95%
accuracy
88%
accuracy

Question 5

WHAT IS
ETHICS?

Answer 5

“[A] set of moral principles/a theory or system of moral values”
(Merriam- Webster)

• “[M]oral principles that govern a person’s behavior or the
  conducting of an activity, the branch of knowledge that deals
  with moral” (Oxford Dictionaries)
• “The field of ethics (or moral philosophy) involves
  systematizing, defending, and recommending concepts of
  right and wrong behavior” (Internet Encyclopedia of
  Philosophy)

Question 6

Ethical frameworks

Answer 6

Consequentialism (J. St. Mill) / utilitarianism - the only thing that matters are consequences of actions and we judge
consequences to be good or bad by looking at whether we harm fewer people than we help

Deontology/Kantian – what matters is the action itself, whether it is good/bad. One should act in accordance with a universal
moral law. How do you know what is a universal law?

Use the categorical imperative: “act only in accordance with that maxim through which you can at the same time
will that it become a universal law”

Virtue ethics (Aristotel) – act as a virtous person would, in this situation + cultivate the right virtues.
Etc.
(The debate is usually between the first two.)

Question 7

Consequentialism (J. St. Mill) / utilitarianism

Answer 7

• the only thing that matters are consequences of actions and we judge
  consequences to be good or bad by looking at whether we harm fewer people than we help

Question 8

Deontology/Kantian

Answer 8

what matters is the action itself, whether it is good/bad. One should act in accordance with a universal
moral law. How do you know what is a universal law?

Question 9

Use the categorical imperative

Answer 9

“act only in accordance with that maxim through which you can at the same time
will that it become a universal law”

Question 10

Virtue ethics (Aristotel)

Answer 10

act as a virtous person would, in this situation + cultivate the right virtues.

Question 11

DATA-DRIVEN
PRICING

Answer 11

Uber uses a dynamic pricing model using data to determine the price

Examples:
* “Uber charged a rider $18,000 to go 11 miles” (The Washington Post)
* Ride was refunded afterwards
* Uber charged the multifold of prices during the Sydney siege (2014)
* Uber charged the multifold of prices during the Hurricane Katrina

What do you think from a utilitarianism point of view/what about as
seen from a Kantian perspective ?

What are the limits of dynamic
pricing? Are there any?

Question 12

WHAT TO PAY ATTENTION TO

Answer 12

1) misrepresenting data
2) misrepresenting insights
3) privacy
4) algorithmic bias
5) misrepresenting algorithms
6) digital divides

Question 13

MAIN SOURCE OF BIAS

Answer 13

We are used to finding and using proxies for various quantities of interest (e.g.
measuring how productive one is by looking at how many parcels they pick in a
given day), but as humans, we are also used to taking such proxies with a grain
of salt, and using common sense to get a holistic view.

Question 14

OK CUPID DATA BREACH

Answer 14

*In 2016, researchers published data of 70,000 OkCupid users—including usernames, political
leanings, drug usage, and intimate sexual details.

‘Some may object to the ethics of gathering and releasing this data. However, all the data found in the dataset are or were already publicly
available, so releasing this dataset merely presents it in a more useful form.’

• Researchers Emil Kirkegaard and Julius Daugbjerg Bjerrekær

*Although the researchers did not release the real names and pictures of the OkCupid users,
critics noted that their identities could easily be uncovered from the details provided—such as fromthe usernames.

Question 15

KEY POINT 1: AWARENESS MATTERS

Answer 15

• ML engineers are often not aware of potential ethical issues

Question 16

KEY POINT 2: ETHICS IS NOT JUST A SELFSERVICE

Answer 16

Dealing with ethics not only for the sake of themselves, ethics can influence
your businesses

Question 17

KEY POINT 3: ETHICAL ISSUES ARE OFTEN
NON-OBVIOUS

Answer 17

• Data science often plays only a single, smaller part in the “equation of evil”
• Difficult to see for the single data scientist
• Only after going through a long chain, the ethical issue becomes visible

Question 18

KEY POINT 4: ETHICS CAN AND SHOULD BE
MANAGED

Answer 18

• Building trust is important to dissipate doubts in ethical integrity