Lecture 1 Flashcards
Why should we care for Data Science & Ethics?
We should care because:
1) It is expected from society
2) There are huge potential risks for both humans (e.g. privacy) and businesses (e.g. reputation)
3) There are multiple potential benefits (e.g. Apple and privacy, bias removed, explainable AI…
Wat are:
Ethics?
Ethics are the moral principles that control or influence a person’s or entity’s behavior or also, what one should do (as opposed to ‘law’ which says what you can do).
What is:
Moral?
Moral is connected with principles of right and wrong behaviour.
What is:
The difference between the utilitarian ethics theory and the deontological one?
The utilitarian ethics theory basically deems any method that has a moral goal as a moral act. The consequence is the main factor to judge morality.
The deontological ethics theory puts emphasis on not taking immoral actions, or actions that ‘feel wrong’.
What influences the equilibrium of data usage?
The equilibrium of data usage is an imaginary spectrum on which on one side, all data is used (e.g. also biased, sensitive…) and on the other side, no data is used.
The equilibrium depends on the context and the weight of either ‘ethics’ or ‘utility’.
For E.g. churn predictions allow for much less ethical concern than CV sorting.
Or predicting when a machine will break down, only raises significant ethical concerns if the operator of the machine is used in the data.
What are:
Issues with a self driving car?
The issues with a self-driving car are mainly about ethical decision making e.g. in case of an emergency, should it drive into a wall and kill a passenger, or drive into a pregnant woman on the other side of the wall?
Why can data be unethical, but an algorithm can’t?
An algorithm can’t be unethical by definition, because it just creates a model on the given input data. E.g. logarithmic regression cannot be unethical (exception; when a model is specifically programmed to look for unethical ds
What is:
Data?
Data are the facts or information, especially when examined and used to find out things or to make decisions
What is:
An Algorithm?
An algorithm is a set of rules that must be followed when solving a particular problem.
What is:
The Prediction or the AI Model?
the prediction (or AI Model) is the decision-making formula, which has been learnt from data by a prediction/AI algorithm*.
What is:
Personal data?
Personal data means any information that can directly or indirectly identify an identified or identifiable natural person (‘data subject’).
E.g. proxis such as name, id-number etc. or one or more factors specific to the physical (mental state, economic background)
What is:
Behavioral data?
Behavioral data is data that provides evidence of a person’s or entity’s actions.
E.g. location data, Facebook likes, online browsing data, payment data
What is:
Sensitive data?
Sensitive data (GDPR, Article 9) is personal data revealing:
1) racial or ethnic origin
2) political opinions
3) religious or philosophical beliefs
4) trade union membership
5) genetic or biometric characteristics for the purpose of uniquely identifying a natural person
6) health
7) a natural person’s sex life or sexual orientation
Are all prohibited
What is:
FAT?
FAT is a model of approaching data science ethics and looks at:
1) Fairness: treating people equally w/o favoritism or discrimination
a) Privacy
b) Discrimination (sensitive data)
2) Accountability is required or expected to justify actions or decisions;
3) Transparency is the ease with which results can be perceived or problems detected
a) Process
b) Explainable AI
