Revision e0c468bc6d3f3d42a2fcb72380d8c4781e576ece (click the page title to view the current version)

AI Ethics

Changes from e0c468bc6d3f3d42a2fcb72380d8c4781e576ece to current

---
title: AI Ethics
categories: session
---

# Reading

Please read [Cervantes et al 2020](https://link.springer.com/article/10.1007/s11948-019-00151-x)

Note, in particular, the following ideas.

- realism and anti-realism as paradigms of morality
    - what is the difference?
    - which one do you believe (most) in?
    - what does the paradigm have to say for the feasibility
      of machine ethics?
- definition (bottom of page 505)
- implicit, explicit, and full ethical agents
    - what is the difference?
    - to what extent can they be reconsiled with the two paradigms?
- top-down, bottom-up, and hybrid approaches
    - what is the difference?
    - to what extent can they be reconsiled with the two paradigms?
- (page 512)  What is meant by a sacred value?
    - is this an important concept in ethics?  in your opinion?
- find at least one example of a proposed AMA system which has been
  implemented and tested
    - which principles does it use?
    - how is it tested?
    - **Note** if you are a slow reader, just pick one at random
      and invest your time to understand that one system

# Briefing (Discussion)

## Ethics in General

Let's disregard AI for the time being.

1.  What do we mean by ethics and morals?
1.  What moral paradigms do we have?  
    - Which different views and approached exist towards ethics?
2.  The idea of a *sacred value* is addressed in the paper
    - what does that mean?
    - do sacred values exist?
    - are they important?
3.  Do human beings always act ethically correct?
    - Is that a problem?

## Ethics in AI

1.  Define an *Artificial Moral Agent* (AMA).
2.  What is the difference between implicit and explicit ethical agents?
3.  Which fundamental approache exist towards implementation of AMA?
4.  Does the choice of moral paradigm matter to the design of AMA?  How?
5.  Moral dilemmata examples.  
    - How do we resolve them as human beings?
    - How do we want an AMA to solve them?

## Principles introduced in the paper

1.  What do we mean by *situationism*?
    - and *empirical*?

# Some more reading

+ [Stanford Encyclopedia of Philosophy](https://plato.stanford.edu/entries/ethics-ai/)
+ Springer [AI & Ethics](https://www.springer.com/journal/43681) journal
+ [Hagendorff 2020](https://link.springer.com/article/10.1007/s11023-020-09517-8)
+ [Bartneck et al (2020)](https://bibsys-almaprimo.hosted.exlibrisgroup.com/permalink/f/19lfmhi/TN_cdi_oapen_doabooks_46811)
+ [Stefan H. Vieweg (ed) 2021](https://bibsys-almaprimo.hosted.exlibrisgroup.com/permalink/f/13q4kuj/BIBSYS_ILS71613727370002201)
+ [EU Guidelines](https://bibsys-almaprimo.hosted.exlibrisgroup.com/permalink/f/19lfmhi/TN_cdi_officepubeu_primary_vtls000465462)