Causality-Preserving Distributed Decision Fabrics for AI-Native Financial Enterprise Architectures
Article Sidebar
Main Article Content
Abstract
Today, financial enterprise architectures are increasingly incorporating artificial intelligence (AI) capabilities into distributed transaction-processing systems, creating a tension between the need for rapid service availability and the need for a causally coherent audit trail of AI-made decisions. In this paper, we marshal research on causal consistency models, causal inference, explainable AI (XAI), and blockchain-based financial infrastructure to propose a conceptual framework called the causality-preserving distributed decision fabric. Causal consistency, which is formalized in causal-memory and session-guarantee models, is explored as a compromise between the high availability of eventual consistency and the coordination cost of linearizability, which is limited by the CAP theorem. Methods for causal inference and model-agnostic explainability, such as Shapley-value attribution and local surrogate modeling, are examined as ways to make automated financial decisions interpretable to those who manage the risks in financial institutions and regulators. Distributed ledger architectures are viewed as an audit trail that can track and order AI-backed decisions between organizational lines. The synthesis suggests that none of the single mechanisms alone meet the simultaneous requirements of availability, explainability, and auditability; thus, a layered architecture of causal-consistency protocols, an explainability layer, and a permissioned ledger is proposed. Comparisons on a conceptual level regarding consistency models and explainability methods, as well as architectural diagrams depicting the proposed layering, are presented. The findings suggest that maintaining causality is a key prerequisite for trustworthy AI-native financial systems, but that interoperability and computational load are major obstacles to implementation.
Article Details
This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.
References
[1] M. Ahamad, G. Neiger, J. E. Burns, P. Kohli, and P. W. Hutto, “Causal memory: Definitions, implementation, and programming,” Distrib. Comput., vol. 9, no. 1, pp. 37–49, 1995, doi: 10.1007/BF01784241.
[2] P. Bailis, A. Fekete, A. Ghodsi, J. M. Hellerstein, and I. Stoica, “HAT, not CAP: Towards highly available transactions,” in 14th Workshop on Hot Topics in Operating Systems (HotOS XIV), Santa Ana Pueblo, NM: USENIX Association, May 2013, doi: 10.1145/2463676.2465279.
[3] S. Gilbert and N. Lynch, “Brewer’s conjecture and the feasibility of consistent, available, partition-tolerant web services,” ACM SIGACT News, vol. 33, no. 2, pp. 51–59, 2002, doi: 10.1145/564585.564601.
[4] D. B. Terry, A. J. Demers, K. Petersen, M. J. Spreitzer, M. M. Theimer, and B. B. Welch, “Session guarantees for weakly consistent replicated data,” in Proceedings of 3rd International Conference on Parallel and Distributed Information Systems, 1994, pp. 140–149, doi: 10.1109/PDIS.1994.331722.
[5] P. Viotti and M. Vukolić, “Consistency in non-transactional distributed storage systems,” ACM Comput. Surv., vol. 49, no. 1, Art. no. 19, 2016, doi: 10.1145/2926965.
[6] J. Pearl, “The seven tools of causal inference, with reflections on machine learning,” Commun. ACM, vol. 62, no. 3, pp. 54–60, 2019, doi: 10.1145/3241036.
[7] Q. Zhao and T. Hastie, “Causal interpretations of black-box models,” J. Bus. Econ. Stat., vol. 39, no. 1, pp. 272–281, Jan. 2021, doi: 10.1080/07350015.2019.1624293.
[8] N. Bussmann, P. Giudici, D. Marinelli, and J. Papenbrock, “Explainable AI in fintech risk management,” Front. Artif. Intell., vol. 3, Art. no. 26, 2020, doi: 10.3389/frai.2020.00026.
[9] D. V. Carvalho, E. M. Pereira, and J. S. Cardoso, “Machine learning interpretability: A survey on methods and metrics,” Electron., vol. 8, no. 8, Art. no. 832, 2019, doi: 10.3390/electronics8080832.
[10] S. Lessmann, B. Baesens, H.-V. Seow, and L. C. Thomas, “Benchmarking state-of-the-art classification algorithms for credit scoring: An update of research,” Eur. J. Oper. Res., vol. 247, no. 1, pp. 124–136, 2015, doi: 10.1016/j.ejor.2015.05.030.
[11] S. M. Lundberg and S.-I. Lee, “A unified approach to interpreting model predictions,” in Advances in Neural Information Processing Systems, 2017, pp. 4765–4774, doi: 10.5555/3295222.3295230.
[12] M. T. Ribeiro, S. Singh, and C. Guestrin, “‘Why should I trust you?’: Explaining the predictions of any classifier,” in Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, in KDD ’16. New York, NY, USA: Association for Computing Machinery, 2016, pp. 1135–1144, doi: 10.1145/2939672.2939778.
[13] O. Ali, M. Ally, P. Clutterbuck, and Y. K. Dwivedi, “The state of play of blockchain technology in the financial services sector: A systematic literature review,” Int. J. Inf. Manage., vol. 54, Art. no. 102199, 2020, doi: 10.1016/j.ijinfomgt.2020.102199.
[14] J. Frizzo-Barker, P. A. Chow-White, P. R. Adams, J. Mentanko, D. Ha, and S. Green, “Blockchain as a disruptive technology for business: A systematic review,” Int. J. Inf. Manage., vol. 51, Art. no. 102029, 2020, doi: 10.1016/j.ijinfomgt.2019.10.014.