Probabilistic AI is a branch of artificial intelligence that models uncertainty explicitly, allowing systems to reason and make predictions even when data is incomplete or noisy. Instead of producing single-point estimates, it generates probability distributions over possible outcomes, capturing both what is known and how confident the model is.
Reference [1] reviewed the research literature on probabilistic AI as applied to finance. Specifically, it followed a rigorous article selection process and ultimately analyzed 62 papers published between 2004 and 2024. The authors pointed out,
In this review, we perform a systematic literature review following a SLR approach to review 62 papers on the topic of probabilistic AI in finance. We examine these papers across dimensions such as model type, output, asset class, and uncertainty type. Additionally, we provide insights into the geographical distribution of research, contributor backgrounds, and the historical development of the field. Our findings suggest that most articles on probabilistic AI claim to enhance point predictions, and few articles have an explicit focus on improving uncertainty estimation within finance. Moreover, probabilistic AI offers valuable capabilities for financial modeling, including non- parametric distribution estimation, separation of uncertainty types, and capturing non-linear dynamics. However, the lack of comprehensive benchmarking and robust evaluations, especially in comparison to traditional models, makes it difficult to assess their true performance.
An important implication of our findings is the need for more interdisciplinary collaboration. Analysis of author backgrounds indicates that research in this area is largely dominated by computer scientists, with relatively limited participation from financial experts. As a result, computer scientists often lack the domain-specific knowledge needed to effectively model financial problems, while financial researchers, despite being better positioned to address such challenges, have seldom adopted probabilistic AI techniques, likely due to technical barriers. This review serves as a starting point for bridging these divides, guiding financial researchers in adopting these methods and helping computer scientists better frame their approaches within the financial context.
In short, the review highlights both the promise and the current limitations of probabilistic AI in finance, particularly the lack of robust benchmarking and systematic evaluation against traditional models. Advancing this field will require stronger interdisciplinary collaboration, where domain expertise from finance and innovation from computer science are combined to produce models that are both technically sound and economically meaningful.
We believe that the conclusion regarding domain knowledge and collaboration applies not only to probabilistic AI but also to deterministic, traditional AI and machine learning in finance.
Let us know what you think in the comments below or in the discussion forum.
References
[1] Sivert Eggen, Tord Johan Espe, Kristoffer Grude, Morten Risstad, Rickard Sandberg, Financial Time Series Uncertainty: A Review of Probabilistic AI Applications, Journal of Economic Surveys, 2025; 00:1–39
Further questions
What's your question? Ask it in the discussion forum
Have an answer to the questions below? Post it here or in the forum
