Back to News

How Generative AI will Impact Financial Institutions

Generative AI (Gen AI) is a rapidly developing technology disrupting business across industries. On March 28th, SIFMA and Deloitte hosted a virtual forum where a panel of specialists discussed use cases, challenges, and risks Gen AI poses for financial institutions. The panel also provided a look into how early adopters are incorporating Gen AI into their businesses. While the technology and potential of Gen AI is generally well understood across the financial industry, its adoption poses various challenges, including cost and implementation complexity, institutions are working to overcome. In this guest post, Deloitte highlights the common themes expressed by participants.

What is Gen AI?

The panel spent the first portion of the forum providing an overview of the various types of Gen AI available today. Gen AI is a form of artificial intelligence that creates content based on foundational models across various modalities such as text, images, audio, code, voice, and video. These foundational models are trained on vast amounts of data and computations to generate initial, knowledge-intensive content based on user prompts. This process of training models and generating content puts companies as both the sources and recipients of AI-generated content. Many large tech firms build and train foundational models and often produce applications that draw upon these models to produce content for users.

Why now?

While Gen AI technology has existed for some time, it has recently gained significant traction. One reason is the proliferation of publicly available applications highlighting its capabilities and possibilities to the broader public. In business, and more specifically among financial institutions, the appetite for use has long been included in strategy discussions. Costs and computing capacity, among other factors, have contributed to the slow adoption and application of Gen AI into business practices. Innovations in machine learning and the cloud technology stack mean the incremental cost of producing knowledge-intensive content such as code, marketing copy, information query, and creative design can fall dramatically.

Gen AI use cases

Many financial institutions are currently looking at how Gen AI can achieve efficiencies across the business. In the near-term, use cases that improve current processes are being explored. On the horizon, use cases aimed at transformational opportunities to scale the business might also be actively pursued and developed.

AI use cases in compliance

The discussion highlighted the use of AI and digital data to remove functional silos and generate broad insights across value chains. However, maximizing AI opportunities necessitates developing enterprise-level AI infrastructure and industrial capabilities. A rising trend is the integration of Gen AI into compliance programs, with applications ranging from money laundering detection to AI-powered issue management and prediction of potential compliance breaches.

AI use cases in banking operations

AI is being used in various operational aspects of banking including digitization of risk assessments, machine learning in reconciliations by reducing reconciliation breaks, and carrying out fraud detection in real-time.

Moreover, AI is also being applied in various front-office procedures. Examples of this include virtual financial advisors, trade and sales support chatbots, and intelligent loan underwriting. These applications have the potential to not only enhance operational efficiency but also greatly improve customer experience and risk management in the banking industry.

Prioritizing use cases: the journey forward

Prioritizing use cases will involve addressing challenges such as data preparation, data quality, data security, and controlling inputs and outputs.

Evaluation criteria will include business value (revenue growth, cost avoidance, speed to market, business impact, risk reduction) and implementation complexity (use case delivery cost, validation effort, ethical implications, data and tech readiness). Quick wins include high value and low complexity use cases.

Emerging regulations and new risks and driving AI risk management enhancement

It was noted that the regulatory landscape is evolving in line with the technological progression. The ability of regulators to provide prescriptive regulation has still not caught up with the rising use of Gen AI tools, however, it is expected the regulators will eventually adapt to the changes. Notably, the SEC proposed rules, which cover technology including AI and mandates certain requirements around identifying and neutralizing conflicts of interest.[i]

Gen AI current risks and challenges to be addressed by AI risk management

The advancement of Gen AI technologies indeed introduces emerging risks and amplifies existing ones, emphasizing the need for an effective AI governance framework. Existing risks that Gen AI can exacerbate include bias, issues with explainability, over-reliance on output, and concerns over confidentiality and privacy. Gen AI brings additional risks such as hallucination (producing inaccurate or misleading results), IP protection and infringement, potential for malicious behavior, token size limits, and increased costs. There’s also a reputational risk associated with these potential issues.

Foundations of a responsible AI risk management framework

Organizations should create and implement AI risk management frameworks that assist them to create and implement AI technologies that are safe and secure, robust and reliable, accountable, responsible, private, transparent and explainable, fair and impartial. Frameworks such as these help organizations to manage AI risks and promote ethical use of AI technologies.

Lessons learned

The rapid evolution of the AI use-environment and the absence of a unified approach in leading AI risk management practices present challenges in designing and implementing AI risk management programs. Lessons learned include the importance of stakeholder alignment, which is often a challenging aspect. It is also beneficial to build on existing processes by adapting established risk management frameworks to cater to AI-specific risks. In the creative environment of AI, managing risk without stifling innovation is crucial. Finally, with AI being a global phenomenon, organizations should consider the complexities of navigating and managing various regulatory requirements across different jurisdictions, which can be a complex task.

This blog contains a summary of the common themes expressed by participants of the SIFMA’s GenAI forum which was held on March 28, 2024. The notes were taken informally and were based on discussions heard and were not validated or confirmed by Deloitte. Deloitte is not, by means of this blog, rendering accounting, business, financial, investment, legal, tax, or other professional advice or services. This blog is not a substitute for such professional advice or services, nor should it be used as a basis for any decision or action that may affect your business. Before making any decision or taking any action that may affect your business, you should consult a qualified professional advisor. Deloitte shall not be responsible for any loss sustained by any person who relies on this blog.

As used in this document, “Deloitte” means Deloitte & Touche LLP, a subsidiary of Deloitte LLP. Please see www.deloitte.com/us/about for a detailed description of our legal structure. Certain services may not be available to attest clients under the rules and regulations of public accounting.

Copyright © 2024 Deloitte Development LLC. All rights reserved

Footnotes

[i] “SEC Proposes New Requirements to Address Risks to Investors From Conflicts of Interest Associated With the Use of Predictive Data Analytics by Broker-Dealers and Investment Advisers,” SEC, July 26, 2023.

Related Resources

More News