Global collaboration outweighs algorithmic complexity in AI-finance research: a negative binomial analysis of citation impact (2015-2025)
This study of 2,694 AI-finance articles (2015–2025) reveals that global collaboration, funding, and open access are the primary drivers of citation impact, while a "Complexity Paradox" penalizes methodologically specialized topics like Deep Learning and interpretable Risk Management, suggesting that audience breadth and novelty outweigh technical sophistication in determining scholarly influence.
Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
In the modern financial world, a quiet revolution is underway. Banks, investment firms, and trading platforms are increasingly relying on artificial intelligence to make decisions, predict market movements, and manage risk. This is not the distant future of science fiction; it is the daily reality of how money moves today. At the heart of this shift is a tension between two competing values. On one side is the drive for technical sophistication, where researchers build incredibly complex computer models capable of finding patterns invisible to the human eye. On the other side is the need for clarity and trust, as financial institutions operate under strict rules that demand they understand how their decisions are made. For decades, scientists have wondered what makes a piece of research truly influential. Does the sheer complexity of the math behind a study guarantee it will be read and cited by others? Or do the most important papers share a different quality, perhaps one related to who wrote them or how broadly their findings can be applied?
A recent study set out to answer these questions by looking at the actual impact of research in the field of artificial intelligence and finance. The researchers gathered a massive collection of nearly 2,700 academic articles published over the last decade, from 2015 to 2025. They examined every paper to see how many times it had been cited by other scholars, a standard measure of how much a piece of work has influenced the scientific community. By analyzing the details of these papers—such as who wrote them, where they were funded, and what specific topics they covered—the team used statistical methods to identify the true drivers of success. They wanted to know if the most technically advanced papers were the most popular, or if something else was at play.
The results revealed a surprising pattern that challenges the common assumption that more complex is always better. The study found that the single strongest predictor of a paper's influence was not the sophistication of its algorithms, but the fact that it was written by a team of authors from different countries. Papers that involved international collaboration were cited nearly twice as often as those written by researchers from a single nation. This suggests that when experts from different parts of the world come together, they create work that resonates with a wider audience and travels further across the globe. Alongside this, the study confirmed that papers with explicit funding support and those available for free to read online also received more attention, though to a lesser degree than international teamwork.
Perhaps the most striking discovery was what the researchers call a "complexity paradox." They found that papers focused on the most advanced and specialized topics actually received fewer citations than expected. Specifically, articles centered on deep learning—a type of artificial intelligence that mimics the human brain with many layers of processing—were cited about 17 percent less than other papers. Even more surprisingly, papers on risk management, which often use clear and understandable methods to predict things like loan defaults, were cited nearly 30 percent less. This finding directly contradicts the idea that the scientific community simply rewards "black box" models that are hard to understand, or conversely, that it strictly favors transparent methods. If the community only cared about transparency, the risk management papers should have been the most popular, but they were not.
Instead, the evidence points to a different explanation: the value of a paper lies in how broadly it can be understood and applied. The topics that received fewer citations tended to be mature, highly specialized, and often focused on specific, local markets. In contrast, the papers that gained the most traction were those that connected to broader audiences, bridging gaps between different fields like economics, computer science, and behavioral studies. The international collaboration premium supports this view, as working across borders naturally forces researchers to make their findings relevant to a wider variety of contexts. The study suggests that in the fast-moving world of financial technology, the most influential work is not necessarily the most mathematically difficult, but the work that speaks to the most people and can be applied in the most places.
This insight carries a significant warning for how we evaluate scientific research. It indicates that simply counting how many times a paper is cited can be misleading if we do not consider the topic and the scope of the work. A paper on a highly specialized, local problem might be excellent work but naturally attract fewer readers than a paper on a new, global trend. The researchers conclude that to truly understand the impact of science in this field, we must look beyond raw numbers and appreciate the role of global connection and general applicability. The future of influential research in finance appears to belong not to those who build the most impenetrable models, but to those who build bridges between different worlds of knowledge.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.