Year: 2025
Type: article
Abstract: Autonomous decision-making in the financial sphere takes a new form due to the blistering development of agentic artificial intelligence (AI). These systems, which are autonomous, goal-directed, and adaptive, are getting more chances in trading, portfolio management, credit scoring, and fraud detection. This survey discusses some of the theoretical foundations that have formed agentic Artificial intelligence, such as decision-theoretic models, reinforcement learning, and belief systems. It has been empirically shown that they are much more effective in dynamic, uncertain scenarios than older AI models, but the problems of transparency, fairness, and robustness persist. The article is strongly critical in evaluating the results of experiments and describes the existing gaps in research studies. Finally, it provides research directions that are necessary to make agentic AI in financial ecosystems safer, interpretable, and regulatory-friendly.
Author: Prajwal Sadananda Nayak
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W4413536117