Projects

Phenomenon-driven AI tools

AI tools for academic literature reviews Current AI tools based on machine learning techniques are mostly limited to providing researchers with basic summaries or article recommendations. We are combining qualitative studies of researchers’ literature review practices with constraint propagation techniques from AI to explore AI tools that enable rich and flexible querying of ideas in the literature.

Frameworks for generating image descriptions Image descriptions generated by computer vision and large language models focus on objectively describing visual content. We draw on qualitative studies of diverse users to frame how AI tools may be designed to support generating image descriptions that also support their dynamic and subjective photographic experiences.

Database for nuruk production and research Organizations’ data are gathered mostly in particular situations, and not in controlled settings ideal for scientific hypothesis testing. Using the setting of nuruk — a Korean fermentation agent with diverse yet untapped biotech applications and little data — we are studying the design of database systems that can integrate data across production and scientific research.
Multisensory archive Current multimodal models are trained on large datasets to leverage statistical patterns between multiple modalities of data (e.g., image-text pairs). Using ethnographic and other qualitative methods, we develop insights into users’ multisensory experiences of data that can be leveraged for building an “archive” that enables meaningful, creative interactions with data.

Problem representation in organizations A long line of business school research views managers as problem-solvers, yet research mostly has focused on how managers search for or predict solutions to given problems. We are integrating models based on the AI idea of “constraint propagation” to research how managers in organizations represent problems from mere symptoms.
Business courses on AI ideas Courses on AI in the business school have been proliferating, but typically are limited to machine learning as a technology. We are developing courses based on viewing AI as a diverse set of ideas (mostly non-ML) that can be used to study organizations as a kind of AI, as well as evaluation frameworks for understanding the use of AI technologies.