Human-AI Symbiosis

Machine learning has become a cornerstone in decision-making across various business landscapes. Recognized for its capacity to discern meaningful patterns within massive datasets, machine learning nonetheless remains susceptible to input manipulation. This is especially true when individuals or entities strategically alter their behaviors or characteristics to achieve a preferred outcome from machine learning models.

A significant focus of my research is dedicated to finding solutions to this challenge. Notably, my ongoing work advocates for a hybrid human-AI approach, leveraging the synergy between human knowledge and AI to enhance the processing of firm disclosures. In such scenarios, adversarial behaviors often surface as managers tactically modify language and information content to soften potential unfavorable investor perceptions.

This research stream aims to devise strategies that counteract these manipulations, ensuring a more authentic and reliable application of machine learning in business contexts.

Related publications:

• Zheng, J., Ng, K. C., Zheng, R., and Tam, K. Y. 2023. “The Effects of Sentiment Evolution in Financial Texts: A Word Embedding Approach,” Journal of Management Information Systems. (41:1), pp. 178-205.