IBM Research @ CHI 2022 - overview
IBM Research participants will be presenting recent advances related to our Human-Centered AI research agenda including 4 full papers, 2 co-organized workshop, 1 co-organized SIG, and 2 workshop papers. Two of the full papers received a Best of CHI Honorable Mention Award. We have a diverse array of contributions focusing on the many different areas of human-computer interaction and data visualization including data practices particularly in data science and artificial intelligence, user understanding of model behavior, research ethics and the development of AI-driven tooling for promoting creative teamwork.
Also check out career opportunities at IBM Research and ways to connect for future opportunities at the end of this post!
Data Practices
While Artificial Intelligence has been identified and used across various domains, incomplete or bad data has been identified as a pain point of Artificial Intelligence usage. One of our papers investigating data science practices through feminism and critical computing received an Honorable Mention Award. The work identified a taxonomy of data silences in data science that lead to incomplete data. The authors provided a detailed analysis of practices that are forgotten in machine learning and propose analytic vocabulary for future work in data science practices [4] . The authors for this paper are also participating in a workshop using the recent taxonomy of data silences to recenter data as an object of interest and to better understand data workers relationships with one another through data [7].
Artificial Intelligence has been touted as way to address high maternal mortality rates in black birthing people through identifying high-risk birthing people to better monitor pregnancies. In the United States, black birthing people are three times more likely to suffer a fatality than their white counterparts. A group of researchers at IBM reviewed the current state of toolkits and digital birthing tailored to black birthing people and found a low prevalence of AI created specifically with Black birthing people and their experiences in mind [2]. The authors make a call to action to reverse the trend of maternal mortality rates in black birthing people.
IBM Researchers have co-organized the "Interrogating Human-Centered Data Science" workshop with other colleagues [5]. The goal of this workshop is to reflect on the HCI work that addresses data science, specifically reflecting on how to humanize data science and take a human-centered approach to address and critique data science.
Human-AI Collaboration
A team of researchers at IBM in collaboration with MIT have created a technique called "Shared Interest" that allows users to compare a models reasoning to that of a human [1]. This paper won the second IBM Best of CHI Honorable Mention Award. With this technique, users are able to sort and rank individual explanations. The authors identified eight recurring patterns in model behavior can help users draw conclusions about the trustworthiness of a model.
IBMers co-organized a workshop on the human understanding of Generative AI systems that can produce new and creative content [6]. The workshop invites participants to discuss the emerging collaboration patterns between creative humans and creative technologies. The aim of the workshop is to develop theories and practices in the domain of generative AI and HCI.
We also have a workshop paper on the intersection of trust and improvisation of human-AI teams presented at the TRAIT workshop that aims to explore appropriate trust and reliance on imperfect AI systems [8] .
Creative Teamwork
Another paper being presented at CHI this year is about a study investigating an AI-based dynamic affective recognition system to support socio-cognitive nonverbal communication in virtual teams. This study find that this tool increases shared understanding and creativity within virtual teams. The study identifies challenges related to transparency and emotional complexity detected by AI [3].
Ethics
We will be participating in the roundtable discussion regarding the ongoing challenge in the HCI research community to develop community norms around ethics. With evolving technology and methods, HCI researchers continue to work on best practices for addressing ethics in research. In a roundtable discussion, members at CHI will be invited to discuss ethical issues both in general research and how they are handled in the SIGCHI paper submission [9].
See below for the full list of works from IBM Research. See you at CHI 2022!
CHI Conference Organizing Committee
Heloisa Candello, David Lindbauer, Pedro Lopez, and Vinoba Vinayagamoorthy serve as co-chairs of CHI 2022 Interactivity.
Jorges Goncalves and Dakuo Wang serve as co-chairs of CHI 2022 Special Interest Groups.
SIGCHI CARES
SIGCHI CARES exists to serve as a resource for those who experience discrimination and/or harassment around our professional events. CARES supports such individuals by allowing them to work with established members of the SIGCHI community, who are approachable and willing to listen and help navigate the SIGCHI and ACM reporting and accountability process.
Celine Latulipe and Michael Muller are co-chairs of SIGCHI CARES. All messages are treated as confidential, sigchi-cares@acm.org
Papers
[1] Angie Boggust, Benjamin Hoover, Arvind Satyanarayan, Hendrik Strobelt. Shared Interest: Measuring Human-AI Alignment to Identify Recurring Patterns in Model Behavior
[2] Morgan Foreman, Jessica Coates, Aminat Adebiyi. The Grand Challenge of Data in Equitably Digitizing Black Maternal Health
[3] Mimi Nguyen, Milad Laly, Bc Kwon, Céline Mougenot. Moody Man: Improving Creative Teamwork Through Dynamic Affective Recognition
[4] Michael Muller, Angelika Strohmayer. Forgetting practices in the data sciences (best paper honorable mention)
Panel
[5] Pranjal Jain, Alex Jordan Blandin, Jacki O'Neill, Mark Perry, Samia Ibtasam, Suleman Shahid, Beni Chugh, David Sullivan, Heloisa Candello, James Pomeroy , Rajat Jain , Robert Dowd , Matt Roach , Matt Jones. Platformisation of Digital Financial Services (DFS): The Journey of DFS in the Global North and Global South
Workshops
[6] Anissa Tanweer, Cecilia Aragon, Michael Muller, Shion Guha, Samir Passi, Gina Neff, Marina Kogan. Interrogating Human-centered Data Science: Taking Stock of Opportunities and Limitations
[7] Michael Muller, Lydia B. Chilton, Anna Kantosalo, Mary Lou Maher, Charles Martin, Greg Walsh. GenAICHI: Generative AI and HCI
[8] Minha Lee, Jaisie Sin, Guy Laban, Mattias Kraus, Leigh Clark, Martin Pocheron, Benjamin R. Cowean, Asbjørn Følstad, Cosmin Munteanu, and Heloisa Candello. Ethics of Conversational User Interfaces.
Workshop papers
[9] Michael Muller, Angelika Strohmayer. Data Silences in Data Preparation: Toward a Taxonomy
[10] Daniel Gonzalez, David Piorkowski, David Mendonca. Behavioral Measures of Trust in Human-Autonomy Teams. CHI 2022 Workshop on Trust and Reliance in Human-AI Teams
[11] Mary Lou Maher, Brian Magerko, Dan Venura, Douglas Fisher, Rogelio E. Cardona-Rivera, Nancy Fulda, John Gero, Minwoo Lee, David Wilson, James C. Kaufman, Maithilee Kunda, Michael Muller, Rachel K.E. Bellamy, Maya Ackerman, Evangelia Chrysikou. A Research Plan for Integrating Generative and Cognitive AI for Human Centered, Explainable Co-Creative AI.
SIG
[12] Heloisa Candello, Adriana S Vivacqua, Pedro Reynolds-Cuéllar, Marisol Wong-Villacres, Laura Sanely Gaytán-Lugo, Adriana Alvarado Garcia. ‘SIGuiendo’ Each Other Steps
[13] Casey Fiesler, Jeff Hancock, Amy Bruckman, Michael Muller, Cosmin Munteanu, Melissa Densmore. Research Ethics in HCI: A SIGCHI Community Discussion
Career opportunities @ IBM Research
