CSCW 2021 conference paper

A CSCW con­fer­ence paper from this year.

Beat­rice Vin­cen­zi, Alex S Tay­lor, Simone Stumpf (2021) Inter­de­pen­dence in Action: Peo­ple with Visu­al Impair­ments and Their Guides Co-Con­sti­tut­ing Com­mon Spaces, Proc. ACM Hum.-Comput. Inter­act. 5(CSCW1), New York, NY, USA: Asso­ci­a­tion for Com­put­ing Machin­ery, pdf, doi:10.1145/3449143

Pri­or work on AI-enabled assis­tive tech­nol­o­gy (AT) for peo­ple with visu­al impair­ments (VI) has treat­ed nav­i­ga­tion large­ly as an inde­pen­dent activ­i­ty. Con­se­quent­ly, much effort has focused on pro­vid­ing indi­vid­ual users with wayfind­ing details about the envi­ron­ment, includ­ing infor­ma­tion on dis­tances, prox­im­i­ty, obsta­cles, and land­marks. How­ev­er, inde­pen­dence is also achieved by peo­ple with VI through inter­act­ing with oth­ers, such as in col­lab­o­ra­tion with sight­ed guides. Draw­ing on the con­cept of inter­de­pen­dence, this research presents a sys­tem­at­ic analy­sis of sight­ed guid­ing part­ner­ships. Using inter­ac­tion analy­sis as our pri­ma­ry mode of data analy­sis, we con­duct­ed an empir­i­cal, qual­i­ta­tive study with 4 cou­ples, each made up of per­son with a vision impair­ment and their sight­ed guide. Our results show how pairs used inter­ac­tion­al resources such as turn-tak­ing and body move­ments to both co-con­sti­tute a com­mon space for nav­i­ga­tion, and repair moments of rup­ture to this space. This work is used to present an exem­plary case of inter­de­pen­dence and draws out impli­ca­tions for design­ing AI-enabled AT that shifts the empha­sis away from inde­pen­dent nav­i­ga­tion, and towards the care­ful­ly coor­di­nat­ed actions between peo­ple nav­i­gat­ing together.

CHI 2021 conference papers

Two papers at the CHI con­fer­ence this year.

Sara Heitlinger, Lara Hous­ton, Alex Tay­lor, Ruth Cat­low (2021) Algo­rith­mic Food Jus­tice: Co-Design­ing More-than-Human Blockchain Futures for the Food Com­mons, Pro­ceed­ings of the 2021 CHI Con­fer­ence on Human Fac­tors in Com­put­ing Sys­tems, New York, NY, USA: Asso­ci­a­tion for Com­put­ing Machin­ery, pdf, doi:10.1145/3411764.3445655

The rela­tion­ships that con­sti­tute the glob­al indus­tri­al food sys­tem tend towards two dom­i­nant val­ues that are cre­at­ing unsus­tain­able social and envi­ron­men­tal inequal­i­ties. The first is a human-cen­tered per­spec­tive on food that priv­i­leges humans over all oth­er species. The sec­ond is a view of food as a com­mod­i­ty to be trad­ed for max­i­mum eco­nom­ic val­ue, reward­ing a small num­ber of share­hold­ers. We present work that explores the unique algo­rith­mic affor­dances of blockchain to cre­ate new types of val­ue exchange and gov­er­nance in the food sys­tem. We describe a project that used role­play with urban agri­cul­tur­al com­mu­ni­ties to co-design blockchain-based food futures and explore the con­di­tions for cre­at­ing a thriv­ing mul­ti­species food com­mons. We dis­cuss how the project helped rethink algo­rith­mic food jus­tice by recon­fig­ur­ing more-than-human val­ues and recon­fig­ur­ing food as more-than-human com­mons. We also dis­cuss some of the chal­lenges and ten­sions aris­ing from these explorations. 

Ceci­ly Mor­ri­son, Edward Cutrell, Mar­tin Grayson, Anja Thieme, Alex Tay­lor, Geert Roumen, Camil­la Long­den, Sebas­t­ian Tschi­atschek, Rita Faia Mar­ques, Abi­gail Sell­en (2021) Social Sense­mak­ing with AI: Design­ing an Open-End­ed AI Expe­ri­ence with a Blind Child, Pro­ceed­ings of the 2021 CHI Con­fer­ence on Human Fac­tors in Com­put­ing Sys­tems, New York, NY, USA: Asso­ci­a­tion for Com­put­ing Machin­ery, pdf, doi:10.1145/3411764.3445290

AI tech­nolo­gies are often used to aid peo­ple in per­form­ing dis­crete tasks with well-defined goals (e.g., recog­nis­ing faces in images). Emerg­ing tech­nolo­gies that pro­vide con­tin­u­ous, real-time infor­ma­tion enable more open-end­ed AI expe­ri­ences. In part­ner­ship with a blind child, we explore the chal­lenges and oppor­tu­ni­ties of design­ing human-AI inter­ac­tion for a sys­tem intend­ed to sup­port social sense­mak­ing. Adopt­ing a research-through-design per­spec­tive, we reflect upon work­ing with the uncer­tain capa­bil­i­ties of AI sys­tems in the design of this expe­ri­ence. We con­tribute: (i) a con­crete exam­ple of an open-end­ed AI sys­tem that enabled a blind child to extend his own capa­bil­i­ties; (ii) an illus­tra­tion of the delta between imag­ined and actu­al use, high­light­ing how capa­bil­i­ties derive from the human-AI inter­ac­tion and not the AI sys­tem alone; and (iii) a dis­cus­sion of design choic­es to craft an ongo­ing human-AI inter­ac­tion that address­es the chal­lenge of uncer­tain out­puts of AI systems. 

Conference papers

I’ve been slow to share papers here, so post­ing about two recent­ly pub­lished papers. With both pub­li­ca­tions it was a absolute joy and priv­i­lege to work with my co-authors.

Cyn­thia L Ben­nett, Daniela K Ros­ner, Alex S Tay­lor (2020) The Care Work of Access, CHI ’20, p. 1–15, New York, NY: ACM Press, pdf, doi:10.1145/3313831.3376568

Cur­rent approach­es to AI and Assis­tive Tech­nol­o­gy (AT) often fore­ground task com­ple­tion over oth­er encoun­ters such as expres­sions of care. Our paper chal­lenges and com­ple­ments such task-com­ple­tion approach­es by attend­ing to the care work of access-the con­tin­u­al affec­tive and emo­tion­al adjust­ments that peo­ple make by notic­ing and attend­ing to one anoth­er. We explore how this work impacts encoun­ters among peo­ple with and with­out vision impair­ments who com­plete tasks togeth­er. We find that bound up in attempts to get things done are con­cerns for one anoth­er and how well peo­ple are doing togeth­er. Read­ing this work through emerg­ing dis­abil­i­ty stud­ies and fem­i­nist STS schol­ar­ship, we account for two impor­tant forms of work that give rise to access: (1) mun­dane attune­ments and (2) non-inno­cent autho­riza­tions. Togeth­er these process­es work as sen­si­tiz­ing con­cepts to help HCI schol­ars account for the ways that intel­li­gent ATs both pro­duce access while some­times sub­vert­ing peo­ple with disabilities.

Jes­si­ca L Feuston, Alex S Tay­lor, Anne Marie Piper (2020) Con­for­mi­ty of Eat­ing Dis­or­ders through Con­tent Mod­er­a­tion, Proc. ACM Hum.-Comput. Inter­act. 4(CSCW1), New York, NY, USA: Asso­ci­a­tion for Com­put­ing Machin­ery, pdf, doi:10.1145/3392845

For indi­vid­u­als with men­tal ill­ness, social media plat­forms are con­sid­ered spaces for shar­ing and con­nec­tion. How­ev­er, not all expres­sions of men­tal ill­ness are treat­ed equal­ly on these plat­forms. Dif­fer­ent aggre­gates of human and tech­ni­cal con­trol are used to report and ban con­tent, accounts, and com­mu­ni­ties. Through two years of dig­i­tal ethnog­ra­phy, includ­ing online obser­va­tion and inter­views, with peo­ple with eat­ing dis­or­ders, we exam­ine the expe­ri­ence of con­tent mod­er­a­tion. We use a con­struc­tivist ground­ed the­o­ry approach to analy­sis that shows how prac­tices of mod­er­a­tion across dif­fer­ent plat­forms have par­tic­u­lar con­se­quences for mem­bers of mar­gin­al­ized groups, who are pres­sured to con­form and com­pelled to resist. Above all, we argue that plat­form mod­er­a­tion is enmeshed with wider process­es of con­for­mi­ty to spe­cif­ic ver­sions of men­tal ill­ness. Prac­tices of mod­er­a­tion reassert cer­tain bod­ies and expe­ri­ences as ‘nor­mal’ and val­ued, while reject­ing oth­ers. At the same time, nav­i­gat­ing and resist­ing these nor­ma­tive pres­sures fur­ther inscribes the mar­gin­al sta­tus of cer­tain indi­vid­u­als. We dis­cuss changes to the ways that plat­forms han­dle con­tent relat­ed to eat­ing dis­or­ders by draw­ing on the con­cept of mul­ti­plic­i­ty to inform design.