Causal and Interpretable AI for Contemporary Art Market Analysis
Supervisor
Suitable for
Abstract
Background Reading:
https://www.engineegroup.com/tcsit/article/view/TCSIT-7-148
https://openreview.net/pdf?id=EO8mTLqDuT
https://openreview.net/pdf?id=EDWTHMVOCj
https://www.amazon.co.uk/Causality-Judea-Pearl/dp/052189560X
This project explores how machine learning and causal inference can enhance transparency and understanding in the contemporary art market. Building on an existing prototype that analyses pricing, provenance, and institutional representation, the project will expand multimodal datasets and refine predictive algorithms to distinguish genuine causal drivers from correlations in art valuation and reputation. By integrating causal discovery methods, explainable AI, and uncertainty estimation, the research aims to produce models that not only predict but also explain how factors such as museum representation, collector networks, and visual characteristics influence market behaviour. The project offers a unique opportunity to contribute to interdisciplinary work at the intersection of AI, econometrics, and the humanities, advancing the interpretability and reliability of algorithmic systems for cultural and financial decision-making.