Roughly a year ago, the cit­izens' coun­cil "AI and Free­dom" met and exchanged views on the role soci­ety could play in AI-research and the sup­port of it. Addi­tion­ally, they dis­cussed the way AI influ­ences our indi­vidu­al and soci­et­al free­dom.

To the RHET AI Cen­ter and espe­cially the Unit 4 (Pub­lic Engage­ment, respons­ible for the idea and organ­isa­tion of the cit­izens' coun­cil "AI and Free­dom"), this coun­cil was an import­ant pro­ject, not only in its exe­cu­tion but also in its pre­par­a­tion and the sub­sequent debriefings.

We inter­viewed a few col­leagues who were involved in vari­ous ways in the ini­tial idea and later exe­cu­tion of the cit­izens' coun­cil about their exper­i­ences regard­ing the pro­ject. By answer­ing 3 ques­tions, they are allow­ing us a glimpse into one cent­ral top­ic each which encap­su­lated them dur­ing the pro­cess, based on their expert­ise and their learn­ings fol­low­ing the course of the pro­ject as well.

Today, our series con­tin­ues with Anika Kais­er (RHET AI Cen­ter Unit 4) who talked to us about epi­stem­ic injustices. 

---

Interview with Anika Kaiser

What are epi­stem­ic injustices and what role do they play in par­ti­cip­at­ory formats?

Anika Kais­er: Epi­stem­ic injustice is a concept coined by Mir­anda Frick­er. It describes why people with cer­tain char­ac­ter­ist­ics sys­tem­at­ic­ally have prob­lems suc­cess­fully par­ti­cip­at­ing in col­lect­ive know­ledge-build­ing pro­cesses. In demo­cra­cies based on the prin­ciple of equal­ity, for example, this ini­tially con­sti­tutes an injustice. How­ever, in addi­tion to being an injustice to cer­tain groups, this injustice also dam­ages entire social or insti­tu­tion­al know­ledge sys­tems: if the know­ledge con­tri­bu­tions of cer­tain groups are sys­tem­at­ic­ally not or poorly 'absorbed' by the sys­tem, this leads to dis­tor­tions and gaps in what a sys­tem agrees upon as its val­id know­ledge base. The per­petu­ation of epi­stem­ic injustice also means that the tools and resources used to gen­er­ate know­ledge in these sys­tems are and remain unsuit­able for artic­u­lat­ing and inter­pret­ing some con­tri­bu­tions to know­ledge.

Among oth­er things, mini-pub­lics are based on the premise that, due to their diverse com­pos­i­tion, they achieve res­ults that bene­fit as many dif­fer­ent social groups as pos­sible, or at least are accep­ted by them. For this reas­on, a great deal of fin­an­cial and organ­iz­a­tion­al effort is inves­ted in ran­domly selec­ted cit­izen for­ums in order to assemble such a diverse advis­ory group. And to ensure that the diverse par­ti­cipants are not only present but can actu­ally con­trib­ute effect­ively with their dif­fer­ent views and exper­i­ences, these formats place increased demands on the qual­ity of dis­course.

Although there are still dif­fer­ences of opin­ion about the exact cri­ter­ia and their pri­or­it­iz­a­tion, there is fun­da­ment­al agree­ment that par­ti­cip­a­tion in joint decision-mak­ing should be free and fair. If epi­stem­ic injustices are effect­ive or poorly addressed in cit­izens' coun­cils, this nat­ur­ally greatly lim­its the qual­ity of discourse—and thus also the res­ults and legit­im­acy of the entire format.

How can epi­stem­ic injustices be addressed? Can epi­stem­ic "justice" be achieved with­in the frame­work of a cit­izens' coun­cil?

Anika Kais­er: I actu­ally try to avoid the term "justice" as much as pos­sible, as it quickly becomes very com­plex in the­or­et­ic­al terms. I use "epi­stem­ic injustice" quite neut­rally as a concept to describe mech­an­isms of exclu­sion for mar­gin­al­ized groups and to make them exam­in­able in prac­tic­al fields. In the the­or­et­ic­al part of my dis­ser­ta­tion, I look at these 'injustices' not primar­ily with the aim of estab­lish­ing justice, but from the per­spect­ive of par­ti­cip­at­ory sci­ence com­mu­nic­a­tion, whose con­cern is, among oth­er things, to max­im­ize the mutu­al gain of know­ledge for sci­ence and soci­ety. My prag­mat­ic goal is there­fore to lim­it epi­stem­ic dam­age in know­ledge sys­tems, wheth­er that be the know­ledge soci­ety as the found­a­tion of our demo­cracy or a know­ledge insti­tu­tion such as the uni­ver­sity.

What can be done in the Cit­izens' Coun­cil is, first of all, to raise aware­ness that bias is some­times unavoid­able. Frick­er sug­gests the "vir­tue of testi­mo­ni­al justice" as a solu­tion, but I have my doubts about the feas­ib­il­ity of this vir­tue. In com­mu­nic­a­tion situ­ations, bias is not only a dis­ad­vant­age, but also con­trib­utes to inform­a­tion pro­cessing and selec­tion. It is very dif­fi­cult to elim­in­ate.

What is nev­er­the­less worth­while is a basic atti­tude in which we first assume that the know­ledge space we know is incom­plete and that we are there­fore depend­ent on addi­tions from oth­ers. I would there­fore argue in favor of choos­ing approaches that relate to the com­mu­nic­a­tion situ­ation rather than cog­nit­ive approaches. The keyword for organ­izers is then above all "agil­ity": con­tinu­ously cre­ate oppor­tun­it­ies for feed­back and adjust­ments to indi­vidu­al needs. In addi­tion, it is import­ant to recog­nize that oppor­tun­it­ies for artic­u­lat­ing and inter­pret­ing exper­i­ences have gen­er­ally been shaped by a dom­in­ant norm. In order to also accom­mod­ate par­ti­cipants who do not con­form to this norm, altern­at­ive chan­nels of com­mu­nic­a­tion can be cre­ated. For example, through images, ges­tures, dance, neo­lo­gisms, draw­ings, sounds, sign lan­guage, etc. How­ever, since this is always a ques­tion of cost and it can be really chal­len­ging to put everything in place as a pre­cau­tion­ary meas­ure, I think an agile, needs-based approach is the obvi­ous choice here too.

What did the Cit­izens' Coun­cil "AI and Free­dom" show you about future approaches to address­ing epi­stem­ic injustice?

Anika Kais­er: In stra­tegic com­mu­nic­a­tion, 'anti­cip­at­ory audi­ence cal­cu­la­tion' plays a very cent­ral role. This means con­sid­er­ing the char­ac­ter­ist­ics and needs of the recip­i­ent in order to tail­or com­mu­nic­a­tion accord­ingly. The the­or­et­ic­al field of epi­stem­ic injustice has made me very aware that such assump­tions guid­ing our actions can also be very pre­ju­diced. Since then, I think I have become some­what more skep­tic­al of my own assump­tions and more open to what my coun­ter­part reveals about them­selves, both in terms of con­tent and form. This applies both in my per­son­al life and in my aca­dem­ic work.

For example, in the Cit­izens' Assembly ques­tion­naires, I asked about vari­ous char­ac­ter­ist­ics, includ­ing rela­tion­ship status. It seemed logic­al to me that people whose part­ners had passed away were no longer in a rela­tion­ship and were there­fore single. That is why I did not offer a sep­ar­ate cat­egory for wid­owed people. When eval­u­at­ing the ques­tion­naires, it became appar­ent that the ques­tion about rela­tion­ship status had been left unanswered in sev­er­al cases and that the item "wid­owed" had been added manu­ally. I can only spec­u­late here that people do not relate the status "wid­owed" solely to the fact that their part­ner died, but that they also feel that this puts them in a spe­cial rela­tion­ship status that dif­fers so greatly from "single" and "mar­ried" that they feel their own situ­ation is incom­pat­ible with these cat­egor­ies.

This example clearly shows that even in sci­ence, inter­pret­ive resources and tools must be examined for their under­ly­ing assump­tions. And indeed, there is evid­ence in my field of research that the loss of loved ones can change one's views in an exclus­ive way that is dif­fi­cult for people without such loss exper­i­ence to under­stand. I there­fore con­clude that, where pos­sible, I should cre­ate space in my research instru­ments for such feed­back on cat­egory errors.

Many thanks to Anika Kais­er for these insights into her research in con­nec­tion with the Cit­izens' Coun­cil "AI and Freedom."

We are also pleased to announce the next inter­view in this series: We inter­viewed Oliv­er Häußler (Uni­ver­sity Com­mu­nic­a­tions, Uni­ver­sity of Tübin­gen) on the top­ic of com­mu­nic­a­tion sur­round­ing the Cit­izens' Coun­cil. Click here for the inter­view with Oliv­er Häußler.

And if you would like to read the first inter­view in this series: Patrick Klü­gel (RHET AI Cen­ter Unit 4, Pub­lic Engage­ment Man­ager at the Uni­ver­sity of Tübin­gen) star­ted the series, and we asked him about expect­a­tion man­age­ment. Click here for the interview.