A joint project between the University of Warwick and Nanyang Technological University, Singapore, is examining how Artificial Intelligence can be made more accessible to small and medium sized enterprises that lack large datasets and expensive digital infrastructure. Led by Dr Yi Ding from Warwick Business School and Professor Vivek Choudhary from NTU, the research focuses on how firms with limited data and technical resources can still use recommendation systems to support digital growth and customer engagement.
Titled Levelling the Artificial Intelligence Playing Field: Evidence on Low Data Recommendation Systems for SME Digital Growth, the project centers on recommendation systems that suggest products or services on digital platforms. These tools are a core part of online commerce, influencing what customers see, click on, and buy. Most existing systems depend on detailed customer histories and sophisticated algorithms that are costly to build and maintain, leaving many smaller firms at a disadvantage compared with larger companies that have stronger data and technical capabilities.
The work builds on a large-scale field experiment conducted with Careem, a ride sharing and delivery platform operating across the Middle East. In that commercial setting, the researchers compared two recommendation approaches. One was a sophisticated, data intensive system that used detailed user histories to infer customer intent and personalize recommendations. The other was a much simpler strategy that required almost no customer data and rotated service options in a largely random way to encourage exploration.
The data intensive personalized system delivered the strongest overall performance, but the simpler low data approach also produced meaningful increases in customer activity and spending. Even without individual user information, the lightweight system was able to prompt engagement and influence behavior. The findings challenge the assumption that effective digital recommendations always require large amounts of data and complex Artificial Intelligence infrastructure.
The project is intended to produce a practical framework that small and medium sized enterprises can adapt for their own digital platforms. The aim is to show when exploration-based strategies work well and when they may outperform more complex data driven systems. The collaboration between Warwick Business School, NTU, and Careem is positioned as the basis for a longer term research program on inclusive digital growth, with an emphasis on helping firms adopt Artificial Intelligence tools responsibly, cost effectively, and at scale.