IBM Research @ IUI 2022     

links

IBM Research @ IUI 2022 - overview


By Zahra Ashktorab, Research Staff Member
 
The ACM IUI Conference on Intelligent User Interfaces is the premiere venue for work that spans across cutting edge Artificial Intelligence and Human Computer Interaction in various domains including psychology, behavioral science, cognitive science, computer graphics, design, the arts. This conference is being held virtually in Helsinki, Finland from March 21 - 25, 2022. 
 
IBM Research participants will be presenting recent advances related to our Human-Centered AI research agenda including 3 full papers, 1 co-organized workshop, 1 co-organized panel, and 2 workshop papers.  This year, many of our papers focused on generative AI across various areas including code translation, co-creative systems, UX modernization, and explainability. 
 
 
Also check out career opportunities at IBM Research and ways to connect for future opportunities at the end of this post!
 
Generative Models, Explainability and Co-creation
The 3rd Workshop on Human-AI Co-Creation with Generative Models is co-organized by three IBMers who work on understanding the incorporation of generative models in various creative domains. Recent advances in generative AI have led to the generation of novel artifacts across different areas including realistic-looking images of faces to antimicrobial peptide sequences that treat diseases to faked videos of prominent business leaders [4].  Understanding how to design, build and evaluate human-AI co-creative systems is an emerging area of interest, particularly for another group of IBMers who are interested in UX modernization. One of the workshop papers explores how to use generative AI to modernize the user experience of legacy applications, a task which in the absence of AI can be manual and arduous [6] . This paper presents opportunities for this growing area of research. Another workshop paper explored the role of generative AI in the role of humans in software engineering [5]. 
 
Explainability is an area that has been explored across discriminative models, but less so across generative AI. A group of IBMers also explored user explainability needs for generative AI, that produces artifacts rather than decisions. In this paper, the authors present XAI features Generative AI that can promote transparency and user understanding [1].  Another paper being presented at the conference explores generative machine learning models in the context of code translation between programming languages. This paper investigated how software engineers utilize AI generated code translations and whether the quality and quantity of AI translations impacts the process [2]. 
 
Machine Learning and Trust
Another paper presented by IBMers addresses the challenges that arise by the widespread use of black-box algorithms making decisions previously entrusted to humans [3]. Interactive Machine Learning (IML) seeks to leverage user feedback to iterate on an ML solution to correct errors and align decisions with those of the users. The authors explore explanation-driven interactive machine learning (XIML) and find that allowing interactivity within interactive machine learning systems causes leads to higher satisfaction, while visual explanations play a less prominent (and somewhat unexpected) role.
 
AI and Health 
IBM Research will also be present on a panel along with other experts from academia and industry discussing the open questions around the design of technology for digital health. The advancements in technology have impacted healthcare with an increasing use of novel devices, sensors and AI algorithms and systems.  The panel will discuss human and machine decision dilemma between health care workers and decision support systems in the medical field [7]
 
See below for the full list of works from IBM Research. See you at virtual IUI 2022! 
 
 
Papers
[1] Jiao Sun, Vera LIAO, Michael Muller, Mayank Agarwal, Stephanie Houde, Kartik Talamadupula, Justin Weisz. Investigating Explainability of Generative Models for Code through Scenario-Based Design
[2]  Justin Weisz, Michael Muller, Steven Ross, Fernando Martinez, Stephanie Houde, Mayank Agarwal, Kartik Talamadupula, John Richards. Better Together? An Evaluation of AI-Supported Code Translation
[3]   Lijie Guo, Elizabeth M. Daly, Oznur Alkan, Massimiliano Mattetti, Owen Cornec, Bart Knijnenburg. Building Trust in Interactive Machine Learning via User Contributed Interpretable Rules
 
Workshops
[4] Justin Weisz, Mary Lou Maher, Hendrik Strobelt, Lydia B. Chilton, David Bau, Werner Geyer. HAI-GEN 2022: 3rd Workshop on Human-AI Co-Creation with Generative Models
 
Workshop papers
[5] Michael Muller, Steven Ross, Stephanie Houde, Mayank Agarwal, Fernando Martinez, John Richards, Kartik Talamadupula, Justin Weisz. Drinking Chai with Your (AI) Programming Partner: A Design Fiction about Generative AI for Software Engineering
[6] Stephanie Houde, Steven Ross, Michael Muller, Mayank Agarwal, Fernando Martinez, John Richards, Kartik Talamadupula, Justin Weisz. Opportunities for Generative AI in UX Modernization
 
Panel 
[7] Ng, Kenney, Nguyen, Dat Quoc, Nguyen, Huyen, Nguyen, Thien, Nguyen, Tuan-Duy Hien, Tran, Thi Ngoc Trang Digital Health: Which Roles for Patients, Professionals and Machines?
 
Career opportunities @ IBM Research
Explore all career opportunities at IBM Research. Connect with us for future opportunities through our IUI form