|Seamless/Seamful HTI Workshop
May 17th and 20th, 2021
Workshop Preparation Checklist
- Submit your one page position paper to the organizers. More information about this is in an email you received.
- Read all of the position papers submitted by your collegues. They can be found here.
- Add three slides to this shared google slide deck to introduce yourself and your work. These will be used during introductions on the first day.
- Sign up for an account on Miro. We will be using this collaborative brainstorming tool during the workshop. You'll need an account to participate.
Thank you! We look forward to seeing you at the workshop.
This NSF sponsored workshop will explore emerging research areas in HCI and AI. The result will be a report we collaboratively author and submit to the NSF that identifies critical research topics, applications, and concerns. Our discussions and explorations will be guided by the themes of seamlessness and seamfulness.
Where are the important seams--the junctures between people and technology, between different modalities, mechanisms, and situations of interaction? How are these seams stitched together? How can we develop new relationships between people and technologies? (When) do we want our interactions with technology to be seamless? (When) do we want to be reminded of the seams?
Seamless/Seamful Human Technology Interaction (HTI) research could involve people who currently work in many different fields, including human-computer interaction, machine learning, data science, haptics, computer graphics and computer vision, cognitive science, privacy and security, critical design, education, human factors, and ethics. Some possible research questions are:
- Seamless Computer Interfaces: Can humans interact with their computers in a more seamless way? When we're typing, computers can help us by filling in words we haven't typed yet. But our interfaces with them are awkward - when we save files, the file manager almost always thinks we want to save it in the wrong folder. Perhaps a draft of a full email response can be written for us along with a proposed folder to save the email in? Imagine the time saved if our interactions with our computers were just a little more seamless.
- Seamful Computer Interfaces: How can we develop technologies and interfaces that give us more agency and control over the technology we use? How much and what kind of agency do we want to retain in our relationships with technology? Do we want to compose our own emails? Organize our own social media feeds? Control our own data and how it is used? How much and what kind of agency do we want to surrender? Who makes these decisions? (When) are seamless technologies convenient and efficient? (When) are they inconvenient, distracting, paternalistic, or exploitive?
- Seamless High-Dimensional Data Analysis: How can we explore data in a way that is more natural than projecting it to 2d and finding clusters, rule mining, histograms or other traditional basic statistical techniques for exploratory data analysis? Perhap even people who can't program could make important discoveries with data if we provide them a way to interact seamlessly with projections of high-dimensional data. This could involve better interactive visualizations, virtual reality to explore projected data spaces, extra dimensions of data being communicated with haptics so that humans could absorb more than a few dimensions of data at once.
- Seamful Interactions for Computational Literacy. How can we make people more cognizant of complex issues in computing, including privacy, security, and potentially biased technologies? Can we intentionally introduce uncomfortable, awkward, or challenging interactions to help people better understand and grapple with the complexity that is often hidden in technological tools?
- Seamless/Seamful Exploration of Machine Learning Model Space: How do we choose the best predictive model for a given task? The space of predictive models that are almost equally good is called the “Rashomon set.” If the Rashomon set is large, domain experts (e.g., doctors, criminal justice experts) need to explore that space to choose the model that works best for the application. What interface would be useful for such users? Can we monitor the way these users interact with the system to determine what part of the model space to "expand" for them to explore more closely?
- Seamless/Seamful Art and Design: Can art and critical design help us develop new ways of engaging with technology and grappling with our ambivalent relationships to it? When do we want technology to fade into the background and when might we want to be (acutely) aware of its presence?
- Seamless/Seamful Communication Awareness: Can AI sensors facilitate better communication between humans? A personal AI monitor could warn you subtly if you seem like you are getting upset before you realize it. Or it could warn you if it seems you are not being clear enough in expressing your ideas using a subtle signal and make a haptic signal as a suggestion for you.
- Seamless User Modeling: Human interaction often builds upon prior knowledge about the interlocutors. We remember previous conversations that not only shape the discussion, but also clarify the expectations of the interlocutors. For a seamless interaction, it is important to incorporate this prior knowledge, which includes previous interactions and implicit clues about the emotional/cognitive state of the others. What is the important user information required for Seamless/Seamful HTI? what are the technologies that have to be included to correctly model the user?
- New Communities, New Seams: How do we meaningfully engage with new (and/or underrepresented) people and communities? Particularly when these communities are disproportionally or negatively impacted by technology? How do we collaborate across disciplinary boundaries? What structures and norms could we adopt to help us develop new (human, social) relationships that impact the design of technology?