Dissecting Corporate Culture with Large Language Models: A Study from Feng Mai of the University of Iowa at the AOM CTO 2026 Symposium

Professor Feng Mai of the University of Iowa presented cutting-edge research on dissecting corporate culture using large language models (LLMs) at the AOM CTO 2026 Symposium. This study combines artificial intelligence technology with the fields of organizational behavior and management to address the problems of time-consuming, labor-intensive, and difficult-to-scale traditional corporate culture research, exploring innovative applications of LLMs in corporate culture analysis, which has significant academic and practical value.

Traditional corporate culture research relies on methods such as qualitative interviews, questionnaires, and case studies. While in-depth, these methods are time-consuming, labor-intensive, and difficult to scale. With the rapid development of artificial intelligence, LLMs have permeated various academic fields. Professor Feng Mai's research represents a cutting-edge exploration of the integration of management and AI technology, aiming to use LLMs to analyze corporate culture characteristics more efficiently and systematically.

1. Text data collection and preprocessing: Collect texts from corporate official websites, employee handbooks, management discussions in financial reports, etc., then clean and preprocess them as input for analysis;
2. LLM application: Identify cultural dimensions (innovation orientation, customer orientation, etc.), evaluate cultural intensity, track cultural changes, and enable cross-organizational comparisons;
3. Validation and calibration: Compare with manual coding results, invite expert evaluations, and test through known cases to ensure result reliability.

Academic contributions: Provide a scalable and replicable analysis tool for organizational culture research, demonstrate the potential of interdisciplinary research between management and computer science, and enable large-scale comparative studies of corporate culture; Practical value: Help enterprises quickly diagnose cultural characteristics, support merger and acquisition integration evaluation, optimize employer brand building, and provide investors with references for evaluating the quality of management teams.

Faces limitations such as text representativeness issues (public texts may not reflect real culture), cultural context understanding difficulties (factors like region/industry affect LLM cognition), challenges in capturing subjectivity (can AI understand the subtleties of culture), and balancing data privacy boundaries (the contradiction between accessing internal documents and privacy protection).

Future directions can include developing multimodal analysis (combining text/images/videos to build comprehensive cultural portraits), real-time cultural monitoring systems, performance prediction models based on cultural characteristics, and cultural design auxiliary tools.

This research represents a microcosm of the digital transformation of management. LLMs have become a new paradigm for understanding complex social phenomena, and their potential needs to be unleashed by combining AI capabilities with human insight. Project link: https://github.com/maifeng/AOM-CTO-2026-Culture-LLMs Related paper: 《Dissecting Corporate Culture with LLMs-Mai-2026.pdf》