Exploratory Customer Discovery on Veblen Goods Using ChatGPT and Agent-based Modeling System for Business Simulation
Document Type
Conference Proceeding
Publication Date
1-1-2025
Abstract
The study of demand in a dynamic and complex market economy has been a point of discussion in observing consumer behavior. Entrepreneurs have struggled with market research as customer discovery can be costly and time-consuming. The concept behind non-intuitive demand curves has increased the complexities of predicting demand. Large language models (LLMs) are observed to be computational models for humans, having the innate capacity to simulate human behavior and decision-making. Through a simulation model and the agent-based modeling system (ABMS) approach, entrepreneurs can utilize LLMs by using their ability for customer discovery. By applying prompting frameworks and chain-of-Thought (CoT), LLMs such as ChatGPT-3.5 can generate customer willingness-To-pay (WTP) without fine-Tuning the model. The study experimented on two business models that pertained to Veblen demand and prompted GPT-3.5 to simulate WTP without any explicit references to "Veblen" to determine its inherent knowledge. The results show that some customer simulations generated with GPT-3.5 appear to form a non-inverted demand curve. Future studies may be considered to understand the LLM's capability of perceiving socio-economic and psycho-emotional behavior regarding product demand and consumption.
Recommended Citation
Alabastro, Z. M., Cordova, R. M., Ilagan, J. B., & Ilagan, J. R. (2025). Exploratory customer discovery on Veblen goods using ChatGPT and agent-based modeling system for business simulation. Procedia Computer Science, 256, 28-35. https://doi.org/10.1016/j.procs.2025.02.092
