Ethics and NLP 2021

This page provides alternative access to the Ethics and NLP resources (Mondays 9:45-11:15 CEST at https://unistuttgart.webex.com/meet/sebastian.pado). If you do not have access to ILIAS, please send me an email (pado@ims) as quickly as possible!

Table to topic assignments:

Date	Topic 1	Topic 2	Topic 3
3.5.	Scientific reading and presenting (Pado)
10.5. Bias	What is Bias? (Kim): Blodgett et al. 2020, Bender & Friedman 2018	Debiasing (Shukla): Hardt 2014, Bolukbasi et al. 2016	More Debiasing (Sinhamapatra): Satell & Abdel-Magied 2020, Lauscher et al. 2020
17.5. Bias, Explanations	Sentiment Analysis (Ashraf): Yu et al. 2012, Kiritchenko & Mohammad 2018	Integrated Explanations (Singh): Goodman & Flaxman 2017, Liu et al. 2019	Post-hoc explanations (Kraemer): Deeks 2019, Ribeiro et al. 2018
31.5. Fake News	Stylometrics (Haque): Juola 2008, Potthast et al. 2018	Political News (Mustafa): Allcott & Gentzkow 2017, Rashkin et al. 2017	Journalism and Fake News (Sorensen): Cohen et al. 2011, Popat et al. 2018
7.6. Fact Checking, Trolling	Explanations for Fact Checking (Schaller): Lim 2018, Atanasova et al. 2020	Trolling (Vostrova): Cheng et al. 2017, Samghabadi et al. 2017	Counteracting Trolling (Hassan): Greenberg 2016, Munger 2017
14.6. Hate Speech	Annotation (Pavolvic-Sukelj): Rahman 2012, Sap et al. 2019	General (Rheinwald): Bleich 2011, Groendahl et al. 2018	Social Media (Quensel): Mueller & Schwarz 2020
21.6. User Profiling, Privacy	Personal Information (Christdoulou): Hovy & Spruit 2016, Benton et al. 2017	Personalization (Olsen): Bright 2008, Grbovic 2018	Search (Ansley): Solove 2007, Zhang et al. 2016
28.6. Privacy, Generation	Text (Casadei): Garcia-Rivadulla 2016, Li et al. 2018	General (Dhyani): Dale 2020, Smiley et al. 2017	Chatbots (Hwang): Schwartz 2019, Vinyals & Le 2015
5.7. Emotions	Motivation and Annotation (Lanfrit): Picard 2004, Schuff et al. 2017	Emotion Generation (Ortiz): Wu 2019, Zhou & Wang 2018	Emotional Chatbots (Yadav): Vincent 2016, Song et al. 2019
12.7. Speech	Multispeaker TTS (Behringer): UK Gov 2019, Jia et al. 2018	Deception (Heitmeier): Levitan et al. 2015, Levitan et al. 2016	Multimodal Information (Cabero): Poropat 2009, Alam & Riccardi 2014
19.7.	Dialogue Assistants and Privacy vs. Emergencies (Wegge): Hoy 2018, Principi et al. 2015	Closing Session

Session of Apr 19: slides including information on how to submit your topic preferences.
Session of Apr 20: slides with announcements.
Please watch the talk by Jonathan Zittrain (1:00-2:30) and take notes about the following points, on which we will base our discussion on the 26th:
1. What is the one topic or statement in the talk that struck you most?
2. What is agenda setting?
3. Zittrain defines AI as 'arcane pervasive tightly coupled adaptive autonomous systems'. Please unpack that definition.
4. Zittrain talks about many problems that involve feedback loops (e.g. voter badge presentation in FB, rider rating in Uber). Are these really AI problems?
5. Two important concepts appears to be fairness and reponsibility. Please define them and outline whether you believe companies need to be fair and responsible.
6. Do you buy the dichotomy between "tools" and "friends" that Zittrain claims, and the technological movement towards "friends"? What consequences does that have?
7. The topic of human involvement comes up several times. Can human involvement solve the problems that were sketched? (Which? How?)
8. At 1:57, Zittrain talks about 'overfitting'. Do you think that this the right term? Why / why not?
9. Why do justifications matter so much?
10. What do you think are reponsibilities of Data Scientists?

Literature archive: literature.tgz