Prof. Anol Bhattacherjee will present the School of IT seminar with a talk entitled, "Two Applications of Artificial Intelligence in Business and Social Research".

Abstract: Artificial intelligence, machine learning, and deep learning is transforming education, businesses, and society for better or for worse. They are also offering new possibilities in how we do research.

In this seminar, Prof Bhattacherjee will discuss two of his current research projects where they used AI/ML to reinvent the way we do qualitative research and clinical mental health screening. In the first project, they analyze regulatory filings by US companies (10-K reports) for two types of innovation activities (exploration and exploitation) using a semi-supervised algorithm that leverages large language models like

GPT-4 and ontologies derived from quantitative (survey) research, with our own extensions in contrastive learning and text augmentation to address the problems of insufficient labelled data and data lacking sufficient diversity. Such automation will allow text coding, which was historically done manually or using rudimental tools like LIWC, to scale up to thousands or millions of documents across diverse contexts such as cybersecurity, customer call transcripts, and social media analysis. In the second project, we use data from a clinical psychiatric evaluation instrument (SCL-10-R) to create explanatory image representations, similar to radiological images like X-rays or CT scans, to map and diagnose ten different mental disorders, using explanatory AI (XAI) approaches using SHAP, LIME, and our own extension called Shapley Radiation, which we validate using clinical data from a mental health clinic in Turkey. Since almost 1 billion people worldwide live with mental disorders, the vast majority of whom are not diagnosed or treated, given one mental health clinician for every 100,000 people in many countries, these mental disorder scans can help vastly scale up the screening of mental disorders and help mental health clinicians manage their exploding workload after the COVID pandemic, diagnose complex multiple disorder cases with overlapping symptoms, and track the progress of their patients over time. The first paper is resubmitted to MIS Quarterly and the second paper to Journal of MIS.