New Research at CAAI: Spring Funded Projects

Fall 2024 marked the launch of a new program at the Center to provide funding for competitive research projects from Booth faculty. Made possible by generous alumni support, the first five faculty projects were funded earlier this year. This Spring, nine additional projects have received support—get a glimpse of what’s on the horizon for AI research below.  

 /1

Building on a project that began in the fall, Daniel Albohn, Principal Researcher at the Roman Family Center for Decision Research, and Alexander Todorov, Walter David “Bud” Fackler Distinguished Service Professor of Behavioral Science, are harnessing generative AI to visualize mental imagery that people associate with certain objects or topics. Starting with human faces, the researchers have been using insights from social psychology and cognitive science to train machine learning models to create photorealistic representations of faces that capture individual representations of attributes like perceived trustworthiness. The latest iteration of the project will extent to other domains such as food to develop images that can inform real-world behavior, such as the perceived healthiness or quality of various foods.

/2

Francesca Bastianello, Assistant Professor of Finance and Liew Family Junior Faculty Fellow, Fama Faculty Fellow, is studying how sell-side analysts reason when valuing stocks. Using large language models (LLMs), the project analyzes the written reasoning in over 2 million equity analyst reports to uncover the mental models (i.e. thought processes) behind analysts' quantitative forecasts. By measuring what analysts choose to focus on, and what simplifying assumptions they adopt when estimating a stock's value, the project sheds light not just on what analysts think, but how they think—bringing new transparency to how they mentally represent firms and how those representations shape investor behavior. Kirill Skobelev, one of the Center’s predoctoral fellows will play an instrumental role in the research.

/3

Also working with equity analyst reports, Leland Bybee, Assistant Professor of Finance at Chicago Booth, is leading an effort to develop a Large Language Model (LLM) to generate expectation data that mimics different subsets of the finance population. Their goal is to assess whether and to what extent LLMs can behave like various subsets of the broader ecosystem of economic and financial analysts.

/4

Led by Giovanni Compiani, Assistant Professor of Marketing, another project will set out to learn how consumers substitute between products as their relative prices change. To this end, the project uses unstructured data (e.g., product descriptions, images, titles, reviews) from a large, global online platform, to capture how similar consumers perceive the different products on the market to be. The project will leverage machine learning models to convert the unstructured data into numeric attributes that can be plugged into an economic model. As information about common products increasingly comes in nonstandard formats, such as through Q&A and review sections of online retailers, this approach aims to mine this data to create a more nuanced understanding of product substitution. This project will be supported by Janani Sekar, one of the Center’s predoctoral fellows.

/5

Kilian Huber, Associate Professor of Economics, Niels Gormsen, Neubauer Family Associate Professor of Finance and Fama Faculty Fellow, and researchers Mary Anderson and Sonali Mishra are analyzing corporate earnings calls to extract price expectation, hurdle rate, and cost of capital data, with the end goal of developing and refining an existing bespoke dataset. The project builds on an earlier initiative with hopes to examine how external factors such as inflation may impact manager perceptions of risk and investment.

/6

Labor economist Anders Humlum, Assistant Professor of Economics at Chicago Booth, is focused on understanding how the adoption of artificial intelligence chatbots like ChatGPT is being adopted within European labor markets. Using large scale surveys of workers from  ranging fields—teachers to programmers to HR professionals—the project aims to capture where chatbots are being folded into workflows and by whom, changes in productivity, the emergence of new AI-empowered tasks, and the overall impact on user earnings and working hours. 

/7

Brian Jabarian, Principal Researcher at the Roman Family Center for Decision Research, is collaborating with leading global companies, innovative startups, and NGOs to investigate the real-world impacts of AI adoption in the economy. He is designing and implementing large-scale natural field experiments to study the impacts of AI in recruitment, healthcare, and agriculture, spanning both developed countries (US, Canada) and developing countries (the Philippines, India, Kenya).

/8

Supported by Kirill Skobelev, Rimmy Tomy, Associate Professor of Accounting and Kathryn and Grant Swick Faculty Scholar, is interested in the role of central banks in climate policy. There is broad debate over whether central banks should actively shape climate policy or focus on ensuring global financial stability. Using ChatGPT and LLMs, the project will sift through pubic comments on a draft proposal of how central banks will engage with climate-related financial risks. By categorizing and analyzing public comments and disclosures, the research team hopes to shed light on whether regulatory “scope creep” is occurring in the financial realm and the influence of public feedback in the process. 
-

Faculty across Booth are engaging in groundbreaking research and innovative applications of AI that have the potential to transform industries and drive social impact. Still in their nascent stages, the spring CAAI-funded projects have the potential to transform industries and drive social impact across sectors. This research would not have been possible without the generous support of the Booth alumni network.

For more coverage of cutting-edge applications of AI, join our mailing list or follow the Center on LinkedIn. 
Faculty – do you have an innovative idea for how to leverage AI for research? Submit your application for research support for the next funding cycle.