RESEARCH PROFILE
Dokyun (DK) Lee is a Kelli Questrom Chair Professor of Information Systems and Computing & Data Sciences at Boston University. His research examines the development, deployment, and impact of artificial intelligence in business and society, with particular emphasis on generative AI, large language models, and unstructured data.
His work studies how AI systems affect firm behavior, consumer behavior, market outcomes, and broader societal consequences, including regulation and governance. This includes empirical and causal analysis of AI reliability, human–AI interaction, and the economic implications of algorithmic systems, with attention to the limitations, failure modes, and unintended consequences that arise when AI technologies are deployed in real-world organizational and legal contexts.
He is the Principal Investigator of the Business Insights through Text (BIT) Lab (www.dkBITLAB.com) and the lead of the Boston University Digital Business Institute Generative AI Lab, where he conducts interdisciplinary research integrating AI, economics, and information systems.
Area of Expertise and Research
DK has published in leading peer-reviewed journals, including Management Science, Information Systems Research, MIS Quarterly, Proceedings of the National Academy of Sciences (PNAS), Science Advances, Nature Scientific Reports, and Journal of Marketing Research, as well as top artificial intelligence venues such as AAAI, AIES, WWW, ACL, and NeuRips Workshop.
DK’s work has received numerous scholarly distinctions, including ISR Best Paper Award (2020), AAAI Award (2021), AMA Don Lehmann Award (2024), Management Science Best Paper Award (2025), 6 finalist distinctions for the Management Science ISR and Marketing Division Best Paper Award, and 13 best-paper awards from prominent conferences (WISE, CIST, ICIS, INFORMS).
DK’s research has been supported by organizations including Adobe, Google, NVIDIA, McKinsey & Company, Bosch Institute, Marketing Science Institute, Net Institute, Prudential Foundation, and MassMutual. His work is frequently consulted in contexts involving AI system impact evaluation, economic impact assessment, regulatory analysis, and disputes concerning the design, deployment, or effects of generative AI technologies, with implications for firms, consumers, and society at large.
DK has done tech consulting for a variety of different firms, including startups, social media companies, financial institutions, and tech companies, in designing and deploying custom deep learning, AI, and Gen AI solutions.
EDITORIAL ROLES
Associate Editor at Management Science
Former Editorial Board at Marketing Science
Former Associate Editor at Information Systems Research
BIO
I hold a Bachelor’s degree in Computer Science from Columbia University (Machine Learning, Artificial Intelligence Focus - Thesis on Human Cognition and Strong AI), a Master’s degree in Statistics (Master's Thesis: Johnson-Lindenstrauss Lemma and its Effect on Supervised Learning) from Yale University and PhD from the Operation, Information and Decisions department of the Wharton School (Thesis: Three Essays in Big Data Consumer Analytics in E-Commerce). I previously worked at many different tech start-ups (mobile, b2b analytics, recommendation web 2.0) and BlackRock as a quantitative software engineer and at Thomson Reuters as a machine learning contractor building a natural language processing engine for financial data. I am an avid fan of Sci-fi series and have watched over 100+ series from the beginning to the end or until their cancellation. I grew up in NY, Queens. AI Evangelist since the 1990s. NLP Engineer since 2006. My friends and colleagues call me DK.
Dokyun (DK) Lee is a Kelli Questrom Chair Professor of Information Systems and Computing & Data Sciences at Boston University. His research examines the development, deployment, and impact of artificial intelligence in business and society, with particular emphasis on generative AI, large language models, and unstructured data.
His work studies how AI systems affect firm behavior, consumer behavior, market outcomes, and broader societal consequences, including regulation and governance. This includes empirical and causal analysis of AI reliability, human–AI interaction, and the economic implications of algorithmic systems, with attention to the limitations, failure modes, and unintended consequences that arise when AI technologies are deployed in real-world organizational and legal contexts.
He is the Principal Investigator of the Business Insights through Text (BIT) Lab (www.dkBITLAB.com) and the lead of the Boston University Digital Business Institute Generative AI Lab, where he conducts interdisciplinary research integrating AI, economics, and information systems.
Area of Expertise and Research
- Generative AI in Business and Society: Economic and societal impact, organizational use, evaluation, governance, and regulatory implications
- Economics of Unstructured Data: Content extraction, value measurement, monetization, and engineering.
- AI Reliability, Validity, and Responsible Uses: Behavioral consistency, robustness, and limits of AI systems.
- Unintended Consequences of AI: Market impacts, societal and regulatory risk
- Customized and Enterprise Human-AI Systems: Design, assessment, and improvements.
DK has published in leading peer-reviewed journals, including Management Science, Information Systems Research, MIS Quarterly, Proceedings of the National Academy of Sciences (PNAS), Science Advances, Nature Scientific Reports, and Journal of Marketing Research, as well as top artificial intelligence venues such as AAAI, AIES, WWW, ACL, and NeuRips Workshop.
DK’s work has received numerous scholarly distinctions, including ISR Best Paper Award (2020), AAAI Award (2021), AMA Don Lehmann Award (2024), Management Science Best Paper Award (2025), 6 finalist distinctions for the Management Science ISR and Marketing Division Best Paper Award, and 13 best-paper awards from prominent conferences (WISE, CIST, ICIS, INFORMS).
DK’s research has been supported by organizations including Adobe, Google, NVIDIA, McKinsey & Company, Bosch Institute, Marketing Science Institute, Net Institute, Prudential Foundation, and MassMutual. His work is frequently consulted in contexts involving AI system impact evaluation, economic impact assessment, regulatory analysis, and disputes concerning the design, deployment, or effects of generative AI technologies, with implications for firms, consumers, and society at large.
DK has done tech consulting for a variety of different firms, including startups, social media companies, financial institutions, and tech companies, in designing and deploying custom deep learning, AI, and Gen AI solutions.
EDITORIAL ROLES
Associate Editor at Management Science
Former Editorial Board at Marketing Science
Former Associate Editor at Information Systems Research
BIO
I hold a Bachelor’s degree in Computer Science from Columbia University (Machine Learning, Artificial Intelligence Focus - Thesis on Human Cognition and Strong AI), a Master’s degree in Statistics (Master's Thesis: Johnson-Lindenstrauss Lemma and its Effect on Supervised Learning) from Yale University and PhD from the Operation, Information and Decisions department of the Wharton School (Thesis: Three Essays in Big Data Consumer Analytics in E-Commerce). I previously worked at many different tech start-ups (mobile, b2b analytics, recommendation web 2.0) and BlackRock as a quantitative software engineer and at Thomson Reuters as a machine learning contractor building a natural language processing engine for financial data. I am an avid fan of Sci-fi series and have watched over 100+ series from the beginning to the end or until their cancellation. I grew up in NY, Queens. AI Evangelist since the 1990s. NLP Engineer since 2006. My friends and colleagues call me DK.
1. Generative AI & Interpretable Machine Learning for Business
Several papers in the impact and application of generative AI in Business:
In addition, widely spread blackbox ML algorithms (e.g., neural nets) do not provide rationale for predictions and are often hard to understand. Consequently, issues involving unintended biases, auditability, trust are emerging in all sectors. In this stream, I develop interpretability-focused algorithms for business applications such as: automatic concept extractor for exploring text data, disentangling factors of technological innovation from patent data, churn prediction model to capture and describe nonlinear consumer-level time series patterns, and a conceptual framework of good explanation for algorithmic transparency, and more with the goal of applying or developing the latest ML models to solve business problems while focusing on interpretability to overcome aforementioned issues.
- Take Caution in Using LLMs as Human Surrogates: Scylla Ex Machina [Presentation Video 2024/09] [Arxiv]
- Generative AI & Human Creativity [Presentation Video]
- Use of Generative AI to Map out Competitive Landscape [Presentation Video]
- Impact of ChatGPT on Knowledge Sharing Websites
- LLM to Nowcast Corporate Risk Event
- Consumer Based Corporate Valuation Via Multi-Tasking Framework
- Guided Diverse Concept Miner [github] [Concept Extractor Algorithm - Not Generative]
In addition, widely spread blackbox ML algorithms (e.g., neural nets) do not provide rationale for predictions and are often hard to understand. Consequently, issues involving unintended biases, auditability, trust are emerging in all sectors. In this stream, I develop interpretability-focused algorithms for business applications such as: automatic concept extractor for exploring text data, disentangling factors of technological innovation from patent data, churn prediction model to capture and describe nonlinear consumer-level time series patterns, and a conceptual framework of good explanation for algorithmic transparency, and more with the goal of applying or developing the latest ML models to solve business problems while focusing on interpretability to overcome aforementioned issues.
2. Economics of Unstructured Data (Content Extraction, Understanding, Engineering, and Marketing)
70-90% of data growth is due to the unstructured data (e.g., texts, images) with much insights still unlocked. In this stream, I measure the economic impact of unstructured data in e-commerce and digital economy. The papers in this stream investigates: what advertising content in social media engage consumers better, what product review content in e-commerce cause conversion, what are factors of technological innovations found in patent text data, what image features are correlated with high demand in lodging market, how do cyberbullying influence Q&A platform and user-contribution, etc. These papers first identify meaningful content by motivating them from theories in consumer behavior, marketing, and economics. Then I apply ML techniques to extract content at scale followed by causal inference to quantify the economic impact of identified content. With these insights, I explore content engineering strategies.
Representative Papers: Facebook Advertising Content Paper, AirBNB and Image Paper, What Review Content Cause Purchase
Representative Papers: Facebook Advertising Content Paper, AirBNB and Image Paper, What Review Content Cause Purchase
3. Unintended Consequence of algorithms and Nudges
Algorithms such as recommender systems have been instrumental in e-commerce settings for increasing sales for retailers and finding new products for consumers. But in this stream of research, I find that popular algorithms such as collaborative filters suffer from unintended consequence of increasing aggregate retailer level sales concentration. On a separate study, different nudges in mobile micro-giving setting is examined.
Journal PAPER LIST