Research
Research at the school of the Intelligent Future of Business can be broadly organized along the following five themes.
AI Enabled Operations
The availability of more data has allowed better predictive and decision models that can enrich operations pipelines. As demonstrated in research by our faculty, this has allowed the detection of early stage disease using algorithms, enabled E-Commerce Retailers to Save Money in their fulfillment operations by carefully modeling and predicting pick failures at stores, and devised split liver transplantation plans that could benefit waiting patients (supported by a gift from Vertex Pharmaceuticals).
Looking forward, by working together with companies, we can advance the design and management of resilient supply chains using these methods.
AI Enabled Marketing
The increasing volume of data and analytical tools has provided a significant increase in understanding and personalizing the customer experience. This is illustrated in our work with Glance, an innovative lock screen content provider in India, which delivers its interactive content cards to their user phones’ lockscreens. We jointly addressed the problem of recommending short-lived high-volume content. We developed new algorithms, validated them on past data, and convinced the business units to run field experiments based on our results. The success of these experiments provided a benefit to the company in improving their customer experience, and to us in publishing our results about the underlying scientific models. The doctoral thesis based on this collaboration also won the prestigious George B. Dantzig Award for the best dissertation in any area of operations research and the management sciences that is innovative and relevant to practice.
Our faculty also work on the economics of deploying AI tools in marketing applications and the key drivers of their success. By working on a combination of devising appropriate signals and tools that AI can enable and linking it with the determinants of good outcomes from deploying such tools, we can advance the development of interactive and customer-centric marketing interventions of the future.
AI Enabled Finance
A key application of data technologies is to improve and automate the detection of potentially fraudulent transactions in a very large data context. Identifying such abnormal transactions among hundreds of thousands of transaction records or spotting money-laundering patterns in large scale transaction data over time are challenges that grow with scale. Our research has shown how this aspect of financial risk management can be tackled by looking for tell-tale patterns in the form of small graph signatures in the data for such abnormal activity, codifying them and looking for them as evidence of abnormality.
Our faculty are also engaged in using AI methods such as natural language processing and machine learning to detect changes in the risk profiles of firms with increased accuracy and explainability. Infusing such AI methods in rigorous financial models can help generate the next generation of financial management tools.
AI Implications for Business
AI has impacted the practice of business as it is adopted in various industries. Our faculty have examined the impact of technology on factory workers, and also grappled with AI Ethics. They have also developed computational methods to predict and identify the correlates of leadership perceptions. Thus, AI has implications in the practice of business beyond the traditional functional silos that are important to investigate as we understand how it helps augment managerial intelligence.
Business Implications for AI
On the other hand, specific business applications of AI also place new requirements on these algorithms for explainability and appropriateness in applications. In this direction, our faculty work on new frameworks to help consumers understand such ML algorithms and examine conditions under which such algorithms can help mitigate racial disparity as a result of their selective adoption. They have also argued that explanations of AI are valuable to those affected by a model’s decisions if they can provide evidence that a past adverse decision was unfair, thus highlighting the factors that contribute to their relative effectiveness.
In addition to suggesting new aspects for explainable AI (XAI) and new requirements in the context of applying them, these applications also sometimes stipulate new AI problems that need to be addressed by AI researchers. Working with our colleagues across campus, we will initiate a virtuous cycle of application driven fundamental AI research arising from business.