People of ACM - Heloisa Candello
January 7, 2025
For someone who is unfamiliar with your work, what impact does the user experience have in fostering ethical AI systems?
Investigating the user experience can unveil the main concerns people have and how they behave when interacting with systems—mainly AI systems. When designers, AI scientists, and software engineering practitioners develop technology, they might have assumptions on how people will interact with what they created. Sometimes they miss what is happening and how their creation can affect the final users. Those issues might not surface without a focused observation on user behaviors.
One of your most downloaded papers in the ACM Digital Library is “ Typefaces and the Perception of Humanness in Natural Language Chatbots,” which was presented at the 2017 ACM CHI Conference. What was the most important insight you and your co-authors found in this research? What is one example of how chatbots have changed in recent years based on user experience considerations?
The principal observation was that individuals frequently felt deceived by chatbots that utilized scripted, human-like writing styles. This sentiment continues to be relevant today, as machines mimicking human behavior can significantly affect trust in these systems. This issue is particularly evident with the current advancements in technology and the ability of models to replicate human traits.
Based on your research, what is an interesting example of how cultural heritage impacts how people use multimedia features in their mobile devices?
In my PhD research, I conducted a user experiment aimed at understanding the use of multimodal features in mobile cultural guides. The focus of the study was a mobile guide prototype that provided information on cultural heritage, specifically sculptures in the United Kingdom. The analysis revealed several relevant findings, including the significance of features such as volume controls on the screen, multitasking functions, and the visibility of hyperlinks. The length and amount of information accessed by users were found to be proportional to their familiarity with the content, language skills, and the time they had available in the city. Visual elements such as pictures aided in learning about the history of the subjects and navigating the tour. Audio was particularly useful when participants needed to examine the details of a sculpture or when light conditions were inadequate for screen visibility. Video content was suitable for reviewing before or after the tour. These findings informed the development of design recommendations for creating effective audiovisual cultural heritage guides.
What is a trend in your field to keep an eye on?
The trend is to create artificial intelligence features—specifically large language models (LLMs)—that embed the guardrails to mitigate harm and take responsibility from the humans in creating detailed and specialized prompts to avoid inappropriate responses from models. It can significantly improve people's experience with generative systems. One of the open challenges that HCI researchers have is identifying the crucial guardrails expected by humans in diverse contexts, cultures, and income levels to inform the creation of representative and responsible LLMs.
Heloisa Caroline de Souza Pereira Candello is a Research Scientist at the Responsible & Inclusive Group at the IBM Research Laboratory in São Paulo, Brazil. She applies her expertise in user research and user experience design to create engaging and ethical AI interactive systems, especially conversational interfaces. One of her primary interests has been in applying design methods to uncover how people use multimedia features in their mobile devices in cultural heritage settings.
She has published multiple papers in prestigious conferences and journals, and received an honorable mention award at CHI 2019. She is also an active volunteer and contributor to the ACM SIGCHI community, where she served as a member of the Volunteer Development Committee for two years and now is a co-chair of the SIGCHI LATAM Committee.
Candello is also an ACM Distinguished Speaker. Her Distinguished Speaker talks include “Artificial Intelligence and Social Impact,” “Generative AI: Design and HCI Perspectives,” and “User Methods and Approaches to Design Conversational User Interfaces.”